[Senate Hearing 115-279]
[From the U.S. Government Publishing Office]
S. Hrg. 115-279
PAVING THE WAY FOR SELF-DRIVING VEHICLES
=======================================================================
HEARING
BEFORE THE
COMMITTEE ON COMMERCE,
SCIENCE, AND TRANSPORTATION
UNITED STATES SENATE
ONE HUNDRED FIFTEENTH CONGRESS
FIRST SESSION
__________
JUNE 14, 2017
__________
Printed for the use of the Committee on Commerce, Science, and
Transportation
[GRAPHIC NOT AVAILABLE IN TIFF FORMAT]
Available online: http://www.govinfo.gov
__________
U.S. GOVERNMENT PUBLISHING OFFICE
30-767 PDF WASHINGTON : 2018
----------------------------------------------------------------------------------------
For sale by the Superintendent of Documents, U.S. Government Publishing Office,
http://bookstore.gpo.gov. For more information, contact the GPO Customer Contact Center,
U.S. Government Publishing Office. Phone 202-512-1800, or 866-512-1800 (toll-free).
E-mail, [email protected].
SENATE COMMITTEE ON COMMERCE, SCIENCE, AND TRANSPORTATION
ONE HUNDRED FIFTEENTH CONGRESS
FIRST SESSION
JOHN THUNE, South Dakota, Chairman
ROGER F. WICKER, Mississippi BILL NELSON, Florida, Ranking
ROY BLUNT, Missouri MARIA CANTWELL, Washington
TED CRUZ, Texas AMY KLOBUCHAR, Minnesota
DEB FISCHER, Nebraska RICHARD BLUMENTHAL, Connecticut
JERRY MORAN, Kansas BRIAN SCHATZ, Hawaii
DAN SULLIVAN, Alaska EDWARD MARKEY, Massachusetts
DEAN HELLER, Nevada CORY BOOKER, New Jersey
JAMES INHOFE, Oklahoma TOM UDALL, New Mexico
MIKE LEE, Utah GARY PETERS, Michigan
RON JOHNSON, Wisconsin TAMMY BALDWIN, Wisconsin
SHELLEY MOORE CAPITO, West Virginia TAMMY DUCKWORTH, Illinois
CORY GARDNER, Colorado MAGGIE HASSAN, New Hampshire
TODD YOUNG, Indiana CATHERINE CORTEZ MASTO, Nevada
Nick Rossi, Staff Director
Adrian Arnakis, Deputy Staff Director
Jason Van Beek, General Counsel
Kim Lipsky, Democratic Staff Director
Chris Day, Democratic Deputy Staff Director
Renae Black, Senior Counsel
C O N T E N T S
----------
Page
Hearing held on June 14, 2017.................................... 1
Statement of Senator Thune....................................... 1
Statement of Senator Peters...................................... 3
Prepared statement........................................... 4
Statement of Senator Moran....................................... 31
Statement of Senator Cortez Masto................................ 33
Statement of Senator Gardner..................................... 36
Statement of Senator Young....................................... 38
Statement of Senator Inhofe...................................... 39
Statement of Senator Hassan...................................... 41
Statement of Senator Blumenthal.................................. 43
Statement of Senator Klobuchar................................... 44
Statement of Senator Cantwell.................................... 46
Statement of Senator Udall....................................... 47
Statement of Senator Markey...................................... 49
Witnesses
Mitch Bainwol, President and Chief Executive Officer, Alliance of
Automobile Manufacturers....................................... 6
Prepared statement........................................... 7
Rob Csongor, Vice President and General Manager of Automotive
Business, NVIDIA Corporation................................... 11
Prepared statement........................................... 13
Colleen Sheehey-Church, National President, Mothers Against Drunk
Driving........................................................ 15
Prepared statement........................................... 16
John M. Maddox, President and Chief Executive Officer, American
Center for Mobility............................................ 18
Prepared statement........................................... 20
Appendix
Jacqueline S. Gillan, President, Advocates for Highway and Auto
Safety, prepared statement..................................... 53
Statement from American Car Rental Association................... 74
Statement from Continental....................................... 76
Statement from Property Casualty Insurers Association of America. 78
Letter dated June 13, 2017 to Hon. John Thune and Hon. Bill
Nelson from Eric Meyhofer, Head of Advanced Technologies Group
(ATG), Uber Technologies, Inc.................................. 79
Article from Reuters dated March 29, 2016, entitled ``Moody's
forecasts U.S. adoption of self-driving cars `several decades'
away '' by Reuters Staff....................................... 83
Letter dated June 13, 2017 to Hon. John Thune and Hon. Bill
Nelson from Electronic Privacy Information Center: Marc
Rotenberg, EPIC President; Caitriona Fitzgerald, EPIC Policy
Director; and Kim Miller, EPIC Policy Fellow................... 84
John Bozzella, President and CEO, Association of Global
Automakers, Inc., prepared statement........................... 87
Statement from the National Association of Mutual Insurance
Companies...................................................... 88
Letter dated June 14, 2017 to Sen. John Thune and Sen. Bill
Nelson from Consumer Watchdog: Harvey Rosenfield, Founder; John
M. Simpson, Privacy Project Director; and Joan Claybrook,
Former President, Public Citizen and Former NHTSA Administrator 90
Response to written question submitted by Hon. Brian Schatz to:
Mitch Bainwol................................................ 114
Response to written question submitted by Hon. Tammy Duckworth
to:
Rob Csongor.................................................. 115
Response to written questions submitted to John M. Maddox by:
Hon. Ron Johnson............................................. 115
Hon. Brian Schatz............................................ 116
Hon. Tammy Duckworth......................................... 117
PAVING THE WAY FOR
SELF-DRIVING VEHICLES
----------
WEDNESDAY, JUNE 14, 2017
U.S. Senate,
Committee on Commerce, Science, and Transportation,
Washington, DC.
The Committee met, pursuant to notice, at 10:05 a.m. in
room SR-253, Russell Senate Office Building, Hon. John Thune,
Chairman of the Committee, presiding.
Present: Senators Thune [presiding], Wicker, Fischer,
Moran, Heller, Inhofe, Capito, Gardner, Young, Cantwell,
Klobuchar, Blumenthal, Markey, Booker, Udall, Peters, Baldwin,
Hassan, and Cortez Masto.
OPENING STATEMENT OF HON. JOHN THUNE,
U.S. SENATOR FROM SOUTH DAKOTA
The Chairman. Before we begin this morning, our thoughts
and prayers are with the members, staff, Capitol Police, and
others who were at the congressional baseball practice this
morning. Capitol Hill is something of a community, and while
details are still coming in, we already know that this
despicable act has affected people that we know and their
families.
I know that many of us in this room have already sent and/
or received text messages or e-mails to and from friends and
family that are checking in. So we obviously want to know what
happened, but the well-being of our friends and colleagues,
especially those wounded, are first in our minds. We're also
very grateful to the Capitol Police, who work diligently to
protect us and, as I understand, saved many lives this morning.
This morning, we're here to talk about self-driving
vehicles. Self-driving vehicles have the potential to make
transportation smarter, safer, and more accessible. A little
over a year ago, we held the first hearing ever in Congress
focusing on this technology and had the opportunity to see some
of these new vehicles in action.
Today, we'll be discussing the great potential of this
technology, but also the numerous policy questions we will need
to address in order to facilitate the testing and safe
deployment of these vehicles without delay. Self-driving
vehicles are poised to bring enormous disruption to our
transportation networks that will improve our lives and our
society. Indeed, the CEO of General Motors has said that she
envisions more change in the auto industry in the next five
years than have occurred in the last 50.
For many Americans, that change can't some soon enough. In
2015, over 35,000 people died in motor vehicle crashes in the
United States, or nearly 100 people per day. That includes car
and truck drivers, passengers, motorcyclists, and pedestrians.
Worldwide, we lose on average 3,500 people per day.
With more than 90 percent of those deaths attributable to
human error, self-driving vehicles--which, after all, can't
fall asleep, use drugs or alcohol, or get distracted by
texting--have the potential to reduce these tragic numbers
dramatically. Drunk driving, in particular, has been a
significant challenge, and I appreciate Mothers Against Drunk
Driving for being here today to discuss this issue.
Self-driving vehicles also have the potential to improve
mobility for many Americans and fundamentally change the way
many of us get around. Offering a new means of accessible
transportation, including for our nation's seniors, and
lessening congestion will improve productivity and efficiency
in all of our lives. Strategy Analytics and Intel predict self-
driving technology will enable a new global passenger economy
worth $7 trillion by the year 2050, representing an incredible
transformation of the world economy.
Self-driving vehicles are no longer a dream of science
fiction. The stepping stones to self-driving vehicles are
already in cars on the road today. Technologies like automatic
emergency braking and lane keeping are proving their worth.
While technology challenges still exist, manufacturers are
becoming increasingly confident that the technology for fully
self-driving vehicles will be ready by 2021, just a few short
years away. In fact, just last week, I rode in a Level 3 Audi
A7 prototype. The company plans to release a Level 3 vehicle to
the public next year.
Manufacturers are asking for regulatory certainty now so
that when the time comes, they will be able to deploy these
vehicles. These great changes are coming fast, and the Federal
Government must be ready to keep pace. As other countries
devote significant attention and effort to stimulating this
technology, strong Federal leadership will be necessary to
maintain our position as a global leader of this innovation.
There are several hurdles to achieving this goal, and Senator
Peters, Ranking Member Nelson, and I are working together on
possible legislative solutions. In fact, just yesterday, we
released principles for bipartisan legislation to guide this
effort.
The transportation laws and regulations currently on the
books did not contemplate the concept of self-driving vehicles.
Current Federal motor vehicle safety standards do not address
automated technologies, and in some cases directly conflict
with them. We are looking for ways to address these conflicts
in dated rules without weakening the important vehicle safety
protections they provide.
We also must be careful to avoid picking winners and losers
in this space. Self-driving vehicles may employ different
technologies, and their deployment may follow varying business
models. So it is important for Congress not to favor one path
before the market figures out what really works best.
While we look for ways to help self-driving vehicles get on
the road quickly, we need to make sure that safety remains our
number one priority. Industry must find ways to show the
technology is safe and reliable to address public skepticism.
The Federal Government must also recognize that it does not
have all the answers. Instead, it must seek outside technology
expertise to begin the hard work of updating existing standards
and setting new rules. As I've said before, government needs to
challenge itself to overcome the traditional 20th century
conception and regulation of a car and a human driver.
AVs will, over time, bring changes to jobs, insurance, law
enforcement, infrastructure, and many other things we cannot
yet foresee. Similar to when the car was first invented, these
challenges are not insurmountable. I remain committed to a
thoughtful discussion with today's panelists, who share my goal
to maintain American innovation leadership and make safety
paramount.
I look forward to working with my colleagues on this
committee as we move forward with potential legislation, and I
now turn to our Ranking Member for the hearing today, Senator
Peters from Michigan, for his opening statement.
STATEMENT OF HON. GARY PETERS,
U.S. SENATOR FROM MICHIGAN
Senator Peters. Thank you, Chairman Thune. It's an honor to
be here, and I appreciate the opportunity to sit in for Senator
Nelson. Unfortunately, Senator Nelson needed to attend to some
other matters and could not be here. But I know he is very much
committed to this issue, and I appreciate him working with you,
Mr. Chairman, to bring this hearing together.
I also appreciate the opportunity to work very closely with
you and your staff, who have been intimately involved in these
sometimes very complex issues, to bring this technology
forward. I'm looking forward to, hopefully, introducing
bipartisan legislation very soon to create a space for this
incredible innovative technology to move forward while also
being very concerned about safety as we move down that path.
As Chairman Thune mentioned, this technology is without
question one of the most transformative technologies to come
out of the auto industry probably since the first car came off
of the assembly line. I believe it is in every way equal to
that innovation, we know what happened when that first
automobile came off of the assembly line. It literally
transformed America. It created the American middle class. It
changed the places where we live, how we get from point A to
point B. It created new industries, creating jobs all
throughout the economy and driving incredible innovation in the
process.
That's going to happen again here at the beginning of the
21st century, and as the Senator representing the great state
of Michigan, the center of the auto industry, it definitely
brings great pride to me and fellow Michiganders. When we think
about this history, it's interesting to think that in the
beginning of the 20th century, it was the automobile that
really propelled incredible innovation in the economy, and,
once again, it's going to be the automobile that moves
innovation forward in the 21st century.
I say that because the technology necessary to move these
vehicles through a very complex environment on the street with
all sorts of potential hazards will mean incredible computer
power, particularly developments in machine learning and
artificial intelligence, and I look forward to hearing from the
panel as to where they see this technology moving beyond
automobiles. What is happening in the auto industry today will
have transformative effects in nearly every industry.
Certainly, safety is paramount. If we can save the lives of the
nearly 40,000 people dying on our highways, that's perhaps the
most significant factor of this great technology.
I will have further comments entered into the record, if I
may, Chairman Thune, because of my unexpected assuming of this
position here with an opening statement. We will have
additional comments put into the record.
[The information referred to follows:]
Prepared Statement of Hon. Gary Peters, U.S. Senator from Michigan
When Henry Ford first installed a moving assembly line in a
Michigan factory in 1913, he couldn't have predicted how this
technology would transform American mobility, help to create the
American middle class, create new industries and jobs, and drive
innovation for the next several decades. Self-driving vehicles have the
potential to be the next transformative technological development in
the automotive industry, and perhaps the most significant since the
assembly line over 100 years ago. As a senator from Michigan, I take
great pride in representing a state that was the center of the
automotive revolution in the early 1900s and which promises to be the
center of innovation in the self-driving vehicle revolution in the
early part of the 2000s.
Self-driving vehicles represent the proverbial ``moon shot'' of the
automotive industry. They have the potential to save tens of thousands
of the more than 35,000 lives lost each year on the road, the vast
majority of which are caused by driver error. They will provide
mobility for those who cannot drive themselves to the grocery store or
to a doctor's appointment, greatly improving the quality of life of
millions of Americans and making them less dependent on assistance for
routine travel. The development of self-driving vehicles will require
increases in computing power, development of advanced machine learning
and artificial intelligence technologies, and increased cybersecurity.
These technologies have the potential to impact a broad number of
industries beyond just the automotive industry, and could lead to new
products and jobs that we haven't even imagined yet.
I believe that it is critical that we ensure American leadership as
the center for automotive innovation in the world. We must act now to
solidify our Nation's position as the world leader in the future of
mobility. Congress can help by making sure that efforts to develop
self-driving vehicles aren't unnecessarily delayed or deterred by
regulations that were designed for a different generation of
automobiles, while remaining primarily focused on the safety of
passengers.
I appreciate the opportunity to work with Senator Thune and his
staff on this matter. We plan to introduce bipartisan legislation that
will allow self-driving technology to continue moving forward while
ensuring that it is done in a safe, responsible, and secure way. The
current regulatory regime requires that carmakers certify that their
vehicles comply with the Federal Motor Vehicle Safety Standards, which
assume the presence of a driver and refer to components like steering
wheels, brake pedals, and rearview mirrors--all of which may be
unnecessary in a self-driving vehicle. Left unchanged, these standards
would make it challenging for manufacturers to test and deploy these
potentially life-saving vehicles, threatening America's position as the
global leader in automotive innovation.
Yesterday, Chairman Thune, Ranking Member Nelson and I released
principles for bipartisan legislation that will address these problems.
This legislation will create a space for innovators to develop new
self-driving vehicle designs while prioritizing innovation and
remaining tech-neutral, allowing for a variety of different forms or
technologies in the vehicles of the future. It will remove barriers to
development that exist today, while working towards a more
comprehensive set of standards as the industry develops and matures. I
greatly appreciate the constructive efforts of the Chairman and Ranking
Member throughout the process of developing this legislation, and I
look forward to its introduction.
A key part of the process of developing new safety standards and
industry norms will be research, development, and testing of self-
driving vehicles in more realistic environments. This will require
proving grounds like the American Center for Mobility in Ypsilanti,
Michigan that can simulate the variety of conditions that these
vehicles will have to deal with, including busy streets, weather
conditions, pedestrians, and more. I believe that public-private
partnerships between the Department of Transportation, industry,
academia, and proving grounds will help ensure continued American
leadership in innovation while speeding the process of developing
safety and testing standards.
America's great innovation engine is poised to lead the world into
a new age of mobility with the development of self-driving vehicles.
Michigan's automotive manufacturers and suppliers, excellent
educational institutions, and now the American Center for Mobility will
be at the center of this disruptive technological revolution. I look
forward to continuing to work with this committee and Chairman Thune to
develop legislation that will help transform mobility and create jobs
in Michigan and around the country.
Senator Peters. I want to take this opportunity to
introduce one witness, and then Chairman Thune will introduce
the other three witnesses.
As I mentioned, my home state of Michigan is definitely
leading the way in innovation, and we have an incredible
concentration of both automotive industry manufacturers and
suppliers. We have engineering talent and research and
development facilities, and I think these are really the focal
point of this technological development.
That's why I'm honored to welcome to the Committee Mr. John
Maddox, who is the President and CEO of the American Center for
Mobility in Ypsilanti, Michigan. This facility is a U.S.
Department of Transportation designated automated vehicle
proving ground and one that will play an essential role in
testing and validating the technologies powering connected and
self-driving cars.
John brings a suite of unique experiences to his leadership
at the American Center for Mobility. John started his career in
automotive safety engineering in NHTSA's Office of Defect
Investigation and later served as Associate Administrator for
Vehicle Safety Research from 2008 to 2012. Originally from
Baltimore, John has strong ties to Michigan, living there with
his wife, Magda, and son, Ben, and working at several
companies, Ford Motor Company, Volkswagen Group, as well as now
at the University of Michigan's Mobility Transformation Center.
John, we're lucky to have you testify today. Because of the
expertise of the American Center for Mobility, we'll definitely
save thousands and thousands of lives every year and
fundamentally change the way we get from place to place in the
future. Thank you for sharing your expertise, and I know the
entire Committee looks forward to your testimony today.
Thank you.
The Chairman. Thank you, Senator Peters, and I'll mention
the other panelists we have before us today: Mitch Bainwol, who
is President and CEO of the Alliance of Automobile
Manufacturers, Incorporated; Mr. Rob Csongor, who is Vice
President and General Manager of automobile business for NVIDIA
Corporation; and Ms. Colleen Sheehey-Church, who is the
National President of Mothers Against Drunk Driving.
We want to welcome you all here as well and appreciate you
taking the time to inform this committee in our deliberations
with your thoughts about this important subject. So we'll start
on my left and your right with Mr. Bainwol, and if you could
confine your oral remarks to five minutes or as close as
possible, that will give us time to ask questions.
So, Mr. Bainwol, please proceed. Welcome.
STATEMENT OF MITCH BAINWOL, PRESIDENT
AND CHIEF EXECUTIVE OFFICER,
ALLIANCE OF AUTOMOBILE MANUFACTURERS
Mr. Bainwol. Thank you, Chairman Thune, Senator Peters, and
members of the Committee. I testify here today on behalf of
auto manufacturers who built 80 percent of the cars on the road
today and who are investing billions of dollars in automation.
I've got to confess I'm a bit distracted by the events that
occurred earlier. Our thoughts and prayers are with the members
and the staff and the security. It puts everything into
perspective.
Let me start by commenting on the bipartisan principles. I
just want to say thank you. It is a wonderful first start. It
frames the debate and the discussion perfectly. So it's a great
start and we're heartened by your leadership.
I'd like to make five broad points as this Committee
continues its work to craft legislation. Point one: four
independent trends are emerging to dramatically reshape
mobility as we know it--increase in automation, connectivity,
ride sharing, and electrification. The move toward autonomy
during this past decade has accelerated, with Driver Assist
that offers important automated functionality, like adaptive
cruise control and active lane keeping. The more consumers
experience Driver Assist, the more they are favorable toward
full automation.
Trend two is connectivity, characterized both by growing
built-in Internet access, as well as communication between cars
and between cars and infrastructure. Trend three is the various
forms of ride sharing. While we think of Uber and Lyft, Car2Go,
Maven, Chariot, and ReachNow, there are many new entrants in
this space, all predicated on the notion that car sharing can
be more efficient use of a high-cost asset than personal
ownership.
And, finally, the fourth trend is electrification.
Electrification has been adopted more slowly than we had
expected and environmentalists hoped, but as range increases
and battery prices fall, the electric power train will become
more competitive. We'll see a tipping point, particularly as
business-owned, self-driving ride sharing fleets emerge.
Point two: the future is here but will take time to fully
kick in. Few debate where we're headed, but there's significant
debate about the length and nature of the journey. The first
automated driving system, so-called SAE Levels 1 and 2, are on
sale today. Level 3 will be for sale, as Senator Thune
indicated, in about a year or so. Introduction of Level 4
probably will begin before 2021, but retail sales to consumers
of Levels 4 and 5 is unlikely to occur before 2025.
Given vehicle cost and how long cars last today, about 20
percent of the cars on the road now were built before 2000. AVs
are not anticipated to be the majority of the fleet for about
three decades, and ubiquity won't come for another 10 years, so
2055, ubiquity. It's equally tricky to predict the percentage
of VMT that will be by personally owned vehicles versus ride
hailing services, but we know this. Change is coming rapidly.
Point three: AVs will usher in a mobility era of profound
social good. Thousands of lives every year can be saved because
technology, while never perfect, can correct for human errors
responsible for so much loss on our roadways. AVs also offer
huge quality of life advantages, access for the disabled and
elderly and time saved for everyone. And, crucially, AVs offer
massive net economic benefits, lower insurance premiums, lower
fuel cost, increased personal productivity, faster commerce,
and better land use.
Yes, AVs will generate disruptions and challenges. No
transition is ever easy, but accelerating this transition
should be our goal.
Point four: the rate of technology innovation is faster
than the rate of regulation and also confuses traditional
regulatory responsibilities. The last NHTSA administrator, Mark
Rosekind, often said government had to be nimble and flexible--
he always said nimble and flexible--because the regulatory
process could not keep up with technological change, and he was
right. The foundation of the Federal AV policy issued last
September was fundamentally sound, relying on guidance rather
than strict rules and seeking to clarify the division of
responsibilities between the Federal Government and the states.
With conventional vehicles, the states regulated the driver
and the feds regulated the vehicle. When the car becomes the
driver, regulatory chaos ensues. A patchwork of different
requirements across states is a recipe for delayed deployment
and delayed realization of the enormity of the benefits that
autonomy offers. There are now 70 bills in 30 states addressing
AVs. So Federal leadership and clear rules are vital.
Fifth and final point: given the enormity of benefit to the
social good, the key question for this Committee is how to
forge public policy that optimizes the safe deployment of these
vehicles. So we have three suggestions toward that end.
First, pass legislation significantly expanding the number
of FMVSS exemptions NHTSA can grant to facilitate more robust
real-world testing, generating the data necessary to refine the
technology before wide scale deployment. Second, pass
legislation clarifying Federal versus State responsibilities to
foster innovation by creating uniformity through a single
Federal framework. And, third, work with DOT to refresh and
modernize existing safety standards to remove obstacles to the
safe deployment of automated vehicles.
The fact that we're here today having this conversation is
tremendously encouraging. Auto makers are eager to work with
you to achieve the remarkable public good within our grasp when
we combine the brilliance of innovation with responsible and
forward-leaning public policy.
I look forward to answering your questions.
[The prepared statement of Mr. Bainwol follows:]
Prepared Statement of Mitch Bainwol, President
and Chief Executive Officer, Alliance of Automobile Manufacturers
Chairman Thune, Ranking Member Nelson, and Members of the Commerce
Committee. Thank you for inviting me to testify today on pathways to
deployment of self-driving vehicles and related technologies. I am here
on behalf of twelve iconic manufacturers who produced 80 percent of the
cars now on American roads and are investing billions of dollars
annually on R&D to improve fuel efficiency and enhance safety. Self-
driving technologies have the potential to do both. I would like to say
from the outset that the Alliance and its members are deeply
appreciative that this Committee, and its House counterpart, have
invested so much time and focus on the various issues implicated by
self-driving cars. We are grateful for your interest in our views and
want to continue being collaborative partners in the process.
Although the automotive sector is highly competitive, we are
unified in recognizing the transformative impact that self-driving
technologies will have on society and the importance of Federal
leadership in removing barriers to their safe development and
deployment.
Unfortunately, we are all familiar with government statistics
regarding highway fatalities: 35,092 people died in traffic crashes in
2015 \1\ an increase over 2014. Preliminary results for 2016 show
another increase. This is a disturbing trend.
---------------------------------------------------------------------------
\1\ NHTSA 2015 Quick Facts
---------------------------------------------------------------------------
The 2015 increase in fatalities is a 7 percent increase from the
prior year. The National Highway Traffic Safety Administration's
(NHTSA) early estimates for calendar year 2016 suggest a possible 10
percent increase. These numbers are concerning and warrant attention,
especially since 94 percent of car crashes are attributable to human
behavior or error (see attached charts). These figures are particularly
relevant to today's hearing and the role that self-driving technologies
can play in possibly reducing overall crashes and fatalities.
I would like to make five broad points to frame the issue and then
close with three recommendations for the Committee to consider as it
works to craft bipartisan legislation to help spur additional
technological and safety advances.
Point 1--Four trends are merging to dramatically reshape mobility
as we know it: increasing automation, connectivity, ride sharing and
electrification. These trends are mutually reinforcing but not mutually
dependent. The move toward autonomy during this past decade has
accelerated significantly--with advanced driver assist systems that
offer important features--like adaptive cruise control and active lane
keeping. Effectively, these technologies have a multiplier impact: the
more consumers experience driver assist systems, the more excited they
become about the prospect of self-driving technologies.
The Alliance has conducted several public opinion surveys that show
the generational shift that is emerging with acceptance of these
technologies. A sample is provided below:
Almost two-thirds (62 percent) of those under 29 years of age are
open to self-driving technology, including 23 percent who view the
technology as ``awesome.'' Only 5 percent of people over 65 years of
age think the technology is ``awesome,'' and almost a third believe
self-driving technologies are a ``terrible idea''--three times higher
than the views of those under 29 years of age. But, importantly,
experience with driver assists has a profound impact on attitudes.
Drivers who have cars with at least two driver assists are dramatically
more favorable (63-35) about autonomy than those who have none (43-54).
Thus, as these technologies make their way into the national fleet,
consumer acceptance will grow materially.
Trend two is connectivity--characterized by growing technological
capabilities that improve the driving experience, vehicle performance
and safety. Trend three is ride sharing--and while we think of
companies like Uber, Lyft, Car2Go, Chariot, Maven and ReachNow to name
a few, there are a huge number of new entrants in this space, all
predicated on the idea that in certain instances car sharing and ride
hailing is a more efficient use of a high cost asset versus personal
ownership. Finally, trend four is electrification. Adoption of
electrification has been slower than some predicted and other experts
hoped--including in California. However, we expect that as range
increases and battery costs fall, EV powertrains will become more
competitive with internal combustion engines. Other coming market
forces, like self-driving ride share fleets, may further spur electric
vehicle deployment. We will see a tipping point--we just do not know
when this will occur.
Point 2--For self-driving technologies, the future is here but will
take a while to be fully realized. Few debate where we are headed.
However, there is significant debate about the length and even nature
of this journey. Keep in mind, even small introductions of self-driving
technologies can reduce fatalities and traffic congestion. The first
driving automation systems--so called SAE Levels 1 and 2--are on sale
today. Introduction of Levels 3, 4, and 5 self-driving technologies, or
Highly Automated Vehicles (HAVs), has yet to begin. Level 3 features,
such as automated driving in freeway traffic jams, are expected to be
introduced soon, perhaps within a year. Level 4 geo-fenced self-driving
vehicles that can only be operated by an Automated Driving System will
probably begin around 2021. But, retail sales to consumers of so-called
Level 5 vehicles that can operate anywhere a person can drive a
conventional vehicle today is unlikely to happen until around 2025 or
after. Given how much vehicles cost and how long they last--more than
20 percent of cars on the road today were produced before 2000 \2\--
vehicles equipped with Level 5 systems will likely not be a majority of
the fleet for three more decades. Ubiquity is not projected to occur
for at least four decades largely due to the fact that over 260 million
light duty vehicles are registered in the U.S. It is also difficult to
predict the percentage of vehicle miles traveled in personally owned
cars versus ride hailing services. But we do know this: change is
coming--and it is coming rapidly.
---------------------------------------------------------------------------
\2\ IHS data compiled by the Auto Alliance
---------------------------------------------------------------------------
Point 3--Self-driving vehicles will usher in a mobility era that
offers profound social benefits. Self-driving technologies will
potentially save thousands of American lives annually, addressing a
large portion of roadway fatalities and crashes associated with human
error. Cars with self-driving features also offer huge quality of life
benefits--access for the disabled and elderly; time saved by being
driven rather than driving so the commuting time can be spent on more
productive activities; and the increased freedom that comes from
quicker trips due to less congestion. Moreover, these technologies
offer massive economic benefits--less congestion, fewer injuries and
medical claims, lower fuel costs, increased personal productivity, and
better land use. The impact on cities may well be enormous. New
communities and municipalities are eager to modernize their mobility
patterns and hunger to learn where new mobility options are headed so
they can begin the infrastructure build-out that could take a decade to
complete. They want to prepare for tomorrow, today. The Commerce,
Science and Transportation Committee has a long history of
understanding the need for and benefits related to uniformity as a
building block for innovation--just look at the railroad, aviation,
telecommunication sectors and the Internet--all of which have spurred
tremendous innovation, social benefits and U.S. leadership.
Point 4--The rate of technology is faster than the rate of
regulation and also confuses traditional regulatory responsibilities.
Self-driving vehicle technologies will generate disruptions and
challenges; no transition is ever easy. However, this is a transition
government and this Committee in particular should seek to accelerate,
because the greater societal good is clear.
The last NHTSA Administrator, Mark Rosekind, was fond of saying
that government must be nimble and flexible because it is difficult for
the regulatory process to keep up with the rapid pace of innovation.
Furthermore, not enough data is in hand to initiate the rulemaking
process to create new standards for self-driving vehicles. If NHTSA
were to prematurely set rules today, it would stifle innovation. The
foundation of the Federal Automated Vehicle Policy (FAVP) that the
Department of Transportation released last September is sound--relying
on overarching guidance rather than rigid rules and seeking to clarify
the division of responsibilities between states and the Federal
Government. Nevertheless, additional Federal leadership is required
here.
With conventional vehicles, the states regulate the driver and the
Federal Government regulates the vehicle. This division of
responsibility still generally makes sense today for self-driving
vehicle technologies, especially since a patchwork of differing safety
and performance standards or other impediments from state to state, and
even city to city, is a recipe for delayed deployment and realization
of the safety and mobility benefits these technologies offer. Take for
instance the fact that so far this year, there have been 70 different
legislative proposals in 30 states that address self-driving vehicles.
As we meet today, the U.S. lacks a critical uniform national framework
to advance these technologies as was established before in the
development of other key innovations.
Federal leadership and clear rules of the road are essential,
especially to underscore NHTSA's authority to issue nationwide safety
and performance regulations for motor vehicles. America is the true
innovation leader in this field. It is in the national interest to
protect that advantage. More importantly, members of the Auto Alliance
share the belief that lives could be lost and that safety improvements
will be delayed without your help.
Point 5--The key question this Committee must ask--is how to use
public policy to optimize the safe deployment of these vehicles and
their promise of social good, while continuing to let innovation spur
economic growth?
Here are three recommendations:
Recommendation 1: Pass legislation significantly expanding the
number and duration of the Federal Motor Vehicle Safety
Standard (FMVSS) exemptions NHTSA can grant under the Safety
Act. There are existing safety standards that serve as direct
barriers to the deployment of self-driving vehicles. Without
providing NHTSA expanded authority to grant exemptions from
these standards, developers will not be able to deploy the
technology at a scale necessary to collect more robust real-
world data to inform future regulatory action.
Recommendation 2: Direct NHTSA to collect the data and
information needed to promptly refresh and modernize the FMVSS
to facilitate the safe deployment of self-driving vehicles. The
Agency should commence such rulemaking without delay after the
necessary data is collected. The existing FMVSS for
conventional vehicles have served the public well. Because they
were intended for vehicles with human drivers, however, they
are ill-suited for vehicles with self-driving technologies.
Alliance members appreciate the need for safety standards and
also believe the process to modernize them for self-driving
vehicles needs to be informed by data generated from increased
exemptions.
Recommendation 3: Pass legislation clarifying Federal versus
state regulatory roles to facilitate innovation and the
expeditious deployment of life-saving self-driving
technologies. This will provide certainty for all stakeholders
in this area and ensure that the United States remains the
leader in self-driving innovation.
We support Federal clarity that will remove or eliminate
impediments to the testing, development, and deployment of
self-driving vehicles--particularly any state laws or
regulations related to the design or performance of these
vehicles. We recognize and continue to support the important
role states play in insurance, licensure, and traffic laws and
enforcement. However, Congress and this Committee should be
aware that state and local laws could still unduly burden or
restrict the use of self-driving vehicles in the future.
Providing Federal clarity on rules governing automated motor
vehicle design, performance and safety does not mean there will
be a vacuum in oversight of the development and deployment of
the technology for both automakers and new entrants. NHTSA has
broad enforcement authority under existing statutes and
regulations to address current and emerging automated safety
technologies. As evidence, look no further than the Enforcement
Bulletin for Emerging Technologies that NHTSA published in
concert with the FAVP last September. That document, which is
still operative, outlines NHTSA's authorities and how they
apply to self-driving technology including software, hardware,
sensors, GPS and vehicle electronics. For example, NHTSA
recently used its extensive investigatory authorities with an
aftermarket self-driving technology company named--Comma.ai--to
ensure it was compliant with regulations before product could
be offered for sale.
The fact that we are all here today having this conversation is
tremendously encouraging. I would like to reiterate the Alliance's and
its members' appreciation of the Committee's work and leadership to
date and indicate our eagerness to continue being a collaborative,
thoughtful partner. The Alliance and its members look forward to
providing constructive feedback on your ideas with a view towards
passing bipartisan legislation. We can achieve remarkable public good
when we marry the brilliance of innovation with responsible and forward
leaning public policy.
Thank you and I look forward to answering your questions.
[GRAPHIC NOT AVAILABLE IN TIFF FORMAT]
______
[GRAPHIC NOT AVAILABLE IN TIFF FORMAT]
The Chairman. Thank you, Mr. Bainwol.
Mr. Csongor?
STATEMENT OF ROB CSONGOR, VICE PRESIDENT
AND GENERAL MANAGER OF AUTOMOTIVE BUSINESS,
NVIDIA CORPORATION
Mr. Csongor. Thank you, Chairman Thune, Senator Nelson,
distinguished of the Committee--I'm sorry, Senator Peters. My
name is Rob Csongor. I'm Vice President and General Manager of
NVIDIA's automotive business.
NVIDIA is one of the world's leading computer technology
companies. We're headquartered in Silicon Valley with more than
10,000 employees across the country and worldwide. I appreciate
your invitation to give testimony today on the important
subject of self-driving. In particular, I'm grateful for the
opportunity to introduce you to the breakthrough work NVIDIA is
doing in artificial intelligence.
Along with hundreds of our partners, we believe AI is the
new computing model, the game changer that makes autonomous
vehicles possible. By understanding how AI works, we can
achieve better regulatory decisions and accelerate our progress
to what we all want, deployment of safe, self-driving vehicles
that will save lives.
NVIDIA's computer innovation is focused at the intersection
of visual processing, high-performance computing, and
artificial intelligence. It's a unique combination that's at
the heart of the world's next-generation computer systems. This
new form of computing is based on our invention of the GPU, or
graphics processing unit, nearly two decades ago. The GPU was
originally designed to power computer graphics, but it has
evolved into a powerful computer brain that processes massive
amounts of data at extraordinary speed.
Ten years ago, researchers began to use GPUs to accelerate
mathematically intense applications, such as mapping the human
genome and predicting weather. More recently, scientists
working in a new field of AI called ``deep learning''
discovered that GPUs are critical to creating algorithms that
enable computers to learn from experience and data, similar to
how the human brain works. In a short period of time, AI
algorithms rapidly outperformed code written manually by
programmers.
As a result, deep learning has become a strategic
imperative across many industries. Consumer services from
companies like Google, Amazon, Microsoft, and Facebook powered
by our technology are now available to millions. In the
healthcare industry, AI is accelerating the search for cancer
cures. For scientists and researchers, NVIDIA delivers
supercomputing solutions used at the Department of Energy, the
Department of Defense, the National Institutes of Health, among
other organizations.
The automotive world is next. A self-driving car is an
immense computational challenge. The car must be able to detect
and perceive objects everywhere around it in motion and in
diverse weather and lighting conditions. The car must determine
its precise position, plan safe paths from one point to
another, and then drive while navigating complex situations
such as construction zones. We simply cannot get there with
conventional programming science. AI technology can solve these
problems.
To this end, NVIDIA has created an open computing platform
comprised of powerful processors which are optimized for AI in
both the car and the data center. In addition, NVIDIA is
developing a full open software stack that the automotive
ecosystem can build around.
Today, we are working with virtually every automaker on
research and development of advanced self-driving vehicles
using AI. Our technology is being used by more than 225
automotive companies worldwide, including Audi, Tesla, Toyota,
Volvo, Mercedes, and others. The Audi A7 that you took a ride
in, Senator Thune, is based on our technology.
We're now at the point where we can create AI systems that
have levels of perception and performance far beyond humans
and, importantly, do not get distracted, fatigued, or impaired.
Much like humans gain knowledge through a lifetime of
experience, AI systems improve over time with additional
training data and testing.
An AI system works by training a deep neural network with
large amounts of data in a data center, monitoring and testing
accuracy. Once validated, the car is updated with new
algorithms over the air, like any modern computer or mobile
device. The car runs the algorithms on real road conditions.
The results are then sent back to the data center, where the
new data can be used to retrain and improve the algorithms. And
then the cycle continues, making the entire fleet better with
each iteration.
Our methodology along with our partners will combine
multiple layers of testing in a data center, on proving
grounds, and on public roads. In addition, leveraging NVIDIA's
experience in visual computing, we can use computer simulation
to accelerate the training and testing process.
So, in conclusion, we believe new regulations are
necessary, but, clearly, there are opportunities to streamline
development and testing. Ideally, we would be able to test cars
and collect diverse data from any state. The patchwork of
different regulations across regions hampers that. It would be
enormously beneficial to have a unified set of regulations
across all states.
It would also be constructive to ensure the standards for
compliance are set correctly. The bar we are comparing against
is a human driver. A system that is significantly safer than a
human driver can save lives once deployed. Conversely,
unrealistic compliance targets runs the risk of costing lives.
And, finally, the deployment of a fleet on real roads
collecting lots of data is the path to achieving safety for the
entire fleet.
Self-driving holds the promise to change our lives. Through
our inventions, our research, and the incredible work of
development partners innovating on our technology, NVIDIA
believes this promise is achievable. We look forward to working
with this committee, the Department of Transportation, NHTSA,
and other groups to ensure the safe deployment of autonomous
vehicles through game-changing technology paired with effective
policy and regulation.
Thank you for the opportunity to tell you of our work.
[The prepared statement of Mr. Csongor follows:]
Prepared Statement of Ron Csongor, Vice President and General Manager
of Automotive Business, NVIDIA Corporation
Thank you, Chairman Thune, Senator Nelson and distinguished members
of the Committee.
I appreciate your invitation to give testimony today on the
important subject of self-driving vehicle technology.
My name is Rob Csongor. I am vice president and general manager of
NVIDIA's Automotive business. NVIDIA is one of the world's leading
computer technology companies, headquartered in Silicon Valley, with
more than 10,000 employees around the country and the world.
NVIDIA is focused on computer innovation at the intersection of
visual processing, high performance computing, and artificial
intelligence or AI--a unique combination at the heart of the world's
next-generation computer systems. Our work has pioneered a supercharged
form of computing relied on by the world's most demanding computer
users including scientists, engineers, designers, and artists.
This form of computing is based on our invention nearly two decades
ago of the GPU or graphics processing unit. The GPU was originally
designed to power computer graphics, the pixels on our computer
screens, but it has evolved into a powerful computer brain that
processes massive amounts of data at extraordinary speed.
Ten years ago, researchers began to use GPUs to accelerate
mathematically intense applications, such as mapping the human genome
and predicting weather. More recently, scientists working in a new
field of AI called deep learning, discovered that GPUs are critical to
creating algorithms that enable computers to learn from experience and
data, similar to how the human brain works. In a short period of time,
AI algorithms rapidly outperformed code written manually by
programmers. As a result, deep learning has become a strategic
imperative for companies across many industries.
Google, Amazon, Microsoft, and Facebook use our AI technology in
thousands of consumer applications. In the healthcare industry AI is
accelerating the search for cancer cures. For scientists and
researchers, NVIDIA delivers supercomputing solutions used at the
Department of Energy, the Department of Defense, and the National
Institutes of Health among other organizations. Our nation's fastest
supercomputer, at Oak Ridge National Lab, is powered by over 18,000
NVIDIA GPUs. And we are working with them to upgrade the system later
this year to our newest processors, which should help the United States
recapture the title of the world's fastest supercomputer.
Self-driving is an immense computational challenge. The car must be
able to detect and perceive many objects, determine its precise
position, plan safe paths from one point to another, and then drive
while navigating complex situations. AI is the new breakthrough in
computing that can solve these problems, and NVIDIA is focused on
delivering this technology.
To this end, NVIDIA has created an open computing platform
comprised of powerful processors in both the car and the data center,
as well as a full, open software stack that carmakers and the ecosystem
are building on.
Today, we are working with virtually every automaker on research
and development of advanced self-driving vehicles using AI. Our
technology is being used by more than 225 automotive companies
worldwide. Audi, Tesla, Toyota, Volvo, and Mercedes have announced they
will deploy vehicles using our technology.
We are now at the point where we can create AI systems that have
levels of perception and performance far beyond humans, and
importantly, do not get distracted, fatigued or impaired.
Much like humans gain knowledge through experience, AI systems
improve over time with additional training data and testing.
The ability to create and test new features and functions, then
securely update the car over the air like a smartphone or personal
computer, enables us to quickly provide cars with safer algorithms and
add more autonomous capabilities once they are proven.
Borrowing from our expertise in visual computing, we can use
computer simulation to test challenging conditions like snow or
blinding sun, as well as potentially hazardous situations without
putting anyone in harms way. Our methodology, along with our partners,
will combine multiple types of testing--in a data center, on proving
grounds, on the road, and in computer simulations.
While we are working with our partners in industry to develop these
technologies, NVIDIA looks forward to working with this Committee, the
Department of Transportation, and NHTSA to ensure the safe deployment
of autonomous vehicles.
With safety as the top priority, we believe new regulations are
necessary. But we also believe there are opportunities to facilitate
development and testing for companies developing these solutions. Safe,
robust AI algorithms are improved through the collection of large
amounts of data. Ideally, we would be able to test fleets across all
states with their diverse driving conditions. A patchwork of different
regulations in different regions hampers development and progress. It
would be enormously beneficial to have a unified set of regulations
across all states.
Together, as industry and government, we will work to safely and
expeditiously bring autonomous vehicles to market and realize their
enormous benefits: saving lives, improving traffic flow, increasing
productivity, and providing mobility to the elderly, the blind and
others who do not have it today. We are committed to this important
mission and to driving the safe development and deployment of
autonomous vehicles in the United States and the world.
Thank you for your time and consideration. I look forward to
answering any questions.
The Chairman. Thank you, Mr. Csongor.
Next up is Ms. Sheehey-Church.
STATEMENT OF COLLEEN SHEEHEY-CHURCH, NATIONAL PRESIDENT,
MOTHERS AGAINST DRUNK DRIVING
Ms. Sheehey-Church. Thank you. Good morning, Chairman
Thune, Senator Peters, and members of the Committee. On behalf
of millions of drunk driving victims across the country, thank
you for the opportunity to testify today and really be a
witness on the critical issue of self-driving cars. You and the
Committee are to be commended for your leadership on highway
safety.
Like so many others before me, I came to Mothers Against
Drunk Driving after my 18-year-old son, Dustin, was killed by a
drunk and drugged driver. Dustin was sober. Dustin was wearing
a seatbelt while riding in the back seat of a car. That vehicle
crashed, went airborne, flew in the air and landed in the
Connecticut River. Dustin could not escape and drowned on July
10, 2004. It was a senseless death and a 100 percent
preventable crime.
Dustin's story is my story. But it does represent thousands
and thousands of stories of victims and survivors in every
state in the country. MADD has changed the culture. In 1980,
over 51,000 people were killed on America's roadways, and over
20,000 people were killed each year due to drunk driving.
Since then, MADD has worked to pass tough drunk driving
laws in all 50 states and at the Federal level. The 21 drinking
age, a national .08 standard, zero tolerance laws for underage
drivers, and ignition interlock laws are just a few of the
things that we have achieved together since 1980.
Total traffic deaths in 2015 were over 35,000. This was an
increase of 7.2 percent over 2014 and represented the largest
increase in 50 years. Unfortunately, preliminary numbers for
2016 show another substantial increase in traffic deaths. That
alone must serve as a wake-up call to the Nation.
According to the National Highway Safety Administration, 94
percent of traffic deaths occur because of human behavior. That
means that 33,000 deaths occurred in 2015 due to the driver.
Many, if not all, of these were preventable. It breaks my heart
to deliver messages like that.
In 2006, MADD launched a new initiative called the Campaign
to Eliminate Drunk Driving, and I know that our goal is
ambitious, but to literally eliminate drunk driving in America.
Our campaign is based on four prongs: number one, to support
our law enforcement; number two, to have ignition interlocks
for all convicted drivers; number three, the development of
advanced vehicle technologies; and, four, grass root support
for all these efforts. That third initiative, that third prong,
is why I'm here today to witness.
We have been pleased to work with this Committee in support
of the Driver Alcohol Detection System for Safety, also known
as DADSS, and progress has been steady, and I would ask this
Committee to help expedite DADSS by encouraging program
partners to have the technology ready for vehicle integration
as soon as possible. But today, I'm here to be a witness on
AVs.
Autonomous vehicles represent the other technology that
actually presents an enormous potential to eliminate drunk
driving and other behavioral traffic deaths. MADD strongly
supports this technology and hopes it will begin as soon and as
safely as possible.
But I need to be clear. MADD is not an expert on the
intricacies of the technology. My panelists are. But we are
absolutely a huge stakeholder. Technology will ultimately be
the way we eliminate drunk driving. Autonomous vehicles are
vital in helping us achieve our goal.
There are some principles that MADD recommends as Congress
and the administration address this issue. Number one, support
Federal regulatory framework. It doesn't make sense for states
to pass laws or to regulate the safety of autonomous vehicles.
It is critical that the Congress and the administration provide
the necessary guidance and regulations to the autonomous
vehicle community in order to allow a very safe national roll-
out.
Number two, support existing state regulatory systems. The
state role in autonomous vehicles is to continue what they do
for conventional vehicles today--titling, registration,
insurance requirements. The states should not regulate AV
safety because they lack the technical expertise to do so.
Number three, we support Level 4 and 5 autonomous
technologies. For MADD's purposes, it is important that these
vehicles achieve the Level 4 and 5 automation to ensure drunk
drivers are separated from driving the vehicle.
Number four, supporting and evaluating the technology as it
evolves over time. NHTSA's Autonomous Vehicle Policy--this data
can be used to enhance the safety benefits of the technology.
In closing, I thank you for the opportunity to testify
before this Committee. The promise of safe self-driving cars is
very exciting, particularly for those of us who have seen the
devastation that human error in driving can bring. I urge the
Committee to hold a hearing on the need to address the increase
in traffic fatalities and work with the Safety Committee and
NHTSA to put these words into actions, and I look forward to
working with this Committee on the issue and answer any
questions later.
[The prepared statement of Ms. Sheehey-Church follows:]
Prepared Statement of Colleen Sheehey-Church, National President,
Mothers Against Drunk Driving
Chairman Thune, Ranking Member Nelson, and Members of the
Committee,
On behalf of millions of drunk driving victims across the country,
thank you for the opportunity to testify before the Senate Commerce
Committee on the critical issue of self-driving cars. We were pleased
to work with you and the Committee on the FAST Act. You and the
Committee are to be commended for your leadership on highway safety.
Like so many others before me, I came to Mothers Against Drunk
Driving after my 18-year-old son Dustin was killed by a drunk and
drugged driver. Dustin was sober and wearing a seatbelt while riding in
the back seat of a car. The vehicle crashed and flipped into the
Connecticut River. Dustin couldn't escape and drowned. It was a
senseless death and a 100 percent preventable crime.
Dustin's story is my story. But it represents thousands and
thousands of stories of victims and survivors in every state in the
country. I travel to those states and talk to those who suffer. . . and
there is one thing that each of them wants . . . NO MORE VICTIMS.
In 1980, a bereaved mother who had lost a child to the needless
crime of drunk driving decided enough was enough. Together with a group
of other moms, she formed Mothers Against Drunk Driving to tell the
country that drunk driving is not a joke, it is a crime. Together with
thousands of other victims, the public and law makers heard our plea
and as a result, laws and attitudes were changed.
Changing the Culture
We changed the culture. In 1980, 51,091 people were killed on
America's roadways and over 20,000 people were killed each year due to
drunk driving. Since then, MADD has worked to pass tough drunk driving
laws in all 50 states and at the Federal level. The 21 minimum drinking
age, a national .08 BAC standard, zero tolerance laws for underage
drivers, and ignition interlock laws are just a few of the things we
have achieved together since 1980.
As a result, drunk driving deaths have been cut in half since our
founding. That's what culture change and tough laws have achieved, but
it's not nearly enough. Thirty-seven years after MADD taught the Nation
it is not acceptable to drink and drive, one third of all traffic
deaths, over 10,000 people, are caused because of drunk drivers.
Total traffic deaths in 2015 were 35,092. This was an increase of
7.2 percent over 2014 and represented the largest increase in 50 years.
Preliminary numbers for 2016 show another substantial increase in
traffic deaths.
This must serve as a wakeup call to the Nation.
According to the National Highway Traffic Safety Administration, 94
percent of all traffic deaths occur because of human error. That means
about 33,000 deaths occurred in 2015 due to the driver. Many if not all
of these were preventable.
Campaign to Eliminate Drunk Driving
In 2006, MADD launched a new initiative called the Campaign to
Eliminate Drunk Driving. Our goal is ambitious, to literally eliminate
drunk driving in America so there are no more victims.
Our Campaign is based on four prongs: support for law enforcement,
ignition interlocks for all convicted drunk drivers, development of
advanced vehicle technologies, and grassroots support for these
efforts.
The third initiative, advanced technology, is why I'm here today.
In 2008, NHTSA and the Automotive Coalition for Traffic Safety
(ACTS) formed an ambitious program called the Driver Alcohol Detection
System for Safety, or DADSS. DADSS seeks to create a passive in-vehicle
alcohol detection system which would prevent an impaired driver at .08
or higher from starting his or her vehicle. The Insurance Institute for
Highway Safety has estimated that the DADSS technology could save 7,000
lives per year. We were pleased to work with you and the Committee
through MAP-21 and the FAST Act to authorize this technology. Progress
has been steady and I would ask this committee to help expedite DADSS
by encouraging program partners to have the technology ready for
vehicle integration by the end of the FAST Act authorization in Fiscal
Year 2020.
Autonomous Vehicles
Autonomous Vehicles represent the other technology that presents an
enormous potential to eliminate drunk driving and other behavioral-
related traffic deaths. MADD strongly supports this technology and
hopes it will begin as soon and as safely as possible.
Let me be clear, MADD is not an expert on the intricacies of this
technology, but we are absolutely a stakeholder. Technology will
ultimately be the way we eliminate drunk driving autonomous vehicles
are vital in helping us achieve our goal.
There are some key principles that MADD recommends as Congress and
the administration address this issue:
Support Federal regulatory framework. It doesn't make sense for
states to pass laws or regulate the safety of autonomous vehicles. It
is important that the playing field be level and states should leave
the self-driving car safety issues to the automotive safety experts at
the federal level. To that end, it is critical that the Congress and/or
the administration provide the necessary guidance and regulations to
the autonomous vehicle community in order to allow a national roll-out
of this technology. Emphasis must be placed on ensuring that the
technology is safe, and that processes are followed to ensure that
rigorous safety standards are followed, and that effective methods for
educating the public on technologies are developed. We believe the DOT/
NHTSA Autonomous Vehicle Policy issued last year is a good beginning
for providing safety guidance to those companies developing Highly
Automated Vehicles. MADD and others in the traffic safety community
look forward to working with the Committee to help ensure that public
safety is of the highest priority, that the development of the
technology continues at a rapid pace, and that future public acceptance
comes to fruition.
Support existing state regulatory system. The state role in
autonomous vehicles is to continue doing what they do for conventional
vehicles today. This includes titling, registration, insurance
requirements, etc. States should not regulate the safety of autonomous
vehicles because they do not have the technical expertise to do so and
their involvement could hinder the technological progress in the
deployment of this life-saving technology.
Support for level four and five autonomous technologies. For MADD's
purposes, it is important that vehicles achieve level four and level
five automation (i.e., vehicles that do not require a human driver in
at least certain environments or under certain conditions). Drunk
driving is arguably the biggest killer on our roadways. Alcohol and
drug impairment have lasting effects, so it is imperative for maximum
safety that the car be able to completely take control and remove the
driver from driving.
Support for evaluating the technology as it evolves. The Autonomous
Vehicle Policy includes provisions for recording and sharing
information on system capabilities and data on events, incidents, and
crashes. These data can be used to enhance the safety benefits of the
technology and should be standardized, anonymous, absent of propriety
information, and made available not only to the Federal Government but
also to states, researchers, and the public.
Closing
In closing, thank you for the opportunity to testify before this
committee. In the last two years, traffic deaths have seen the largest
increases in 50 years. This should alarm us all. While great progress
has been made in reducing traffic deaths, over 35,000 people died in
2015. This is unacceptable. The promise of safe, self-driving cars is
very exciting, particularly for those of us who have seen the
devastation that human error and driving can bring. I want to add,
however, that we must not sit back and wait for technology to solve the
public health epidemic on our roads. There are actions we can take
today, this month, this year that will make an enormous impact and save
lives. I urge the Committee to hold a hearing on the need to more
aggressively address the increase in traffic fatalities, and work with
the safety community and NHTSA to put words into action.
Longer-term, Autonomous Vehicle technology provides the opportunity
to eliminate traffic deaths, including drunk driving fatalities. There
is a major role for the Federal Government to play in ensuring that
these vehicles are safe. It is also important to create a national
regulatory framework to ensure that the technology can move forward and
build public support.
MADD will continue to strongly support the safe development of this
technology and work to build public acceptance for the adoption of
autonomous vehicles so that people understand the safety benefit of
these cars with regard to behavioral safety.
Thank you again for including MADD in this panel. I look forward to
working with this committee on this issue and am happy to answer any
questions you might have.
The Chairman. Thank you, Ms. Sheehey-Church, for sharing a
deeply personal and painful experience and your effective
advocacy.
Ms. Sheehey-Church. You're welcome.
The Chairman. Mr. Maddox?
STATEMENT OF JOHN M. MADDOX, PRESIDENT AND CHIEF EXECUTIVE
OFFICER, AMERICAN CENTER FOR MOBILITY
Mr. Maddox. Good morning. Thank you, Chairman Thune,
Ranking Member Peters, for your interest in automated vehicles
and your leadership on this issue, and I certainly thank you
for the opportunity to speak in front of you today about key
steps for creating a much safer and more efficient
transportation system for the United States of America.
The American Center for Mobility is a nonprofit public-
private partnership. That's a mouthful. We are building a
world-leading facility for innovation, testing and product
development, and to act as a national proving ground for
automated vehicles in Ypsilanti, Michigan.
First and foremost, we are focused on safety, on public
safety, including testing and validation of these technologies,
including vehicles and infrastructure. In addition to testing,
our mission is to accelerate the development of standards,
voluntary and Federal regulation, that is, and to provide
educational opportunities for the workforce of the future, for
STEM K-12 programs, and the public, in general.
Our past transportation system has served us well over the
last 100 years but is showing signs of strain. Every year, we
endure a national tragedy of tens of thousands of Americans,
individuals, losing their lives on our roads in the United
States. That's the numerical equivalent of 10 September 11
attacks or seven Iraq wars every year on our soil. In many
cities, we see ever-increasing congestion with accompanying
loss of productive time, loss of time, wasted energy, and
unnecessary greenhouse gas emissions. Largely, we have come to
accept these as the status quo.
It is clear we need a significant change in our
transportation system going forward, and safety must be
everyone's first focus. My colleagues and the Chairman have
cited the statistics. I won't do that again. But it is clear
that this technology can help by reducing human error. In
addition, transportation really is the lifeblood of our economy
and our society. The ability to move people efficiently and
goods is critical to the social and economic well-being of the
U.S. and will help us remain competitive with other
international economies.
Because of this promising safety potential and economic
benefit, the U.S. should focus clearly and specifically on
developing and deploying connected automated vehicle technology
in a responsibly expedient manner. AV technology is being
developed very rapidly by industry, as we know, but also aided
by key frameworks and research from government and academia.
Other countries, including China and the EU, are currently
working on HAV programs, very significant programs.
Our voluntary standards have proven to be a key early step
in the development of almost every safety technology. These
voluntary standards often form the basis for Federal FMVSS;
therefore, it is critical to accelerate voluntary standards in
order to ensure safe deployment of HAVs. SAE International--the
Society of Automotive Engineers International--has begun to
promulgate basic standards, but many additional voluntary
standards are needed immediately, including scenario catalogs,
test procedures, mapping, labeling, cybersecurity, to name a
few. The complexity of AV technology will require innovative
thinking for testing simulation, validation, and certification
tools to support these vehicles and infrastructure.
Voluntary standards must also be accelerated for purposes
of avoiding a patchwork, as even a small number of differing or
conflicting regulations would significantly inhibit the
development of AV technology. Consumers' interests would not be
served if they could not operate their vehicle or share a
vehicle across state lines. Certainly, differing standards
would put the U.S. in a compromised position compared to other
countries and regions around the world.
It may be prudent to consider the adoption of a Federal
FMVSS framework standard that establishes key foundational
requirements, such as definitions, manufacturer identification,
reporting processes, et cetera; however, it is too early to
promulgate detailed vehicle performance or equipment standards
as the needed technical requirements to do so are not
sufficiently developed, and, currently, there is no agreement
within the technical and stakeholder community on the nature or
specifics of such requirements.
U.S. DOT's Automated Vehicle Proving Ground Program, AVPG
for short, is important to the safe development of HAVs. The
AVPG program with 10 designated sites across the United States
provides key infrastructure and framework for the development
of AV tools, products, standards, data sharing in a rapid and
collaborative fashion. These proving grounds can form the
backbone of the Federal Government's framework for supporting
HAVs. This program will be a critical asset to ensure the focus
on safety in HAVs while ensuring the U.S. remains competitive.
The need for HAV data collection will continue after
initial deployment, and exemptions are important to allow that
data generation. According to U.S. DOT, a number of existing
FMVSS contain requirements that conflict with or do not allow
basic HAV design. In order to enable early product development
and regulatory data gathering and to enable HAVs, in general,
it's important that Congress and the U.S. DOT review that
authority and implement a solution for exempting compliance of
a sufficiently substantial number of vehicles. It does remain
imperative for HAVs that the petitioner demonstrate that the
exempted vehicle provides an equivalent level or better of
safety through sufficient data and analysis.
In closing, I'd like to offer three recommendations. One,
that U.S. DOT be authorized and funded to create collaborative
voluntary industry standards to support safe deployment. Two,
the AVPG program should be funded as important infrastructure
for HAVs. And, lastly, Congress should consider revising
NHTSA's exemption authority and ensuring appropriate FTE head
count to implement these key programs.
I appreciate this opportunity and look forward to your
questions. Thank you.
[The prepared statement of Mr. Maddox follows:]
Prepared Statement of John M. Maddox, President
and Chief Executive Officer, American Center for Mobility
On behalf of the American Center for Mobility, it is an honor to
provide testimony about the future of automated vehicle (AV) and
transportation technology, and key steps for creating a much safer and
more efficient transportation system.
This testimony will focus on the need to maximize the benefits of
new transportation technologies for the United States, while ensuring
safe deployment on public roads. It will explain how voluntary
standards inform Federal regulation, and how these can be accelerated
through coordinated industry and government collaboration, including
the USDOT Automated Vehicle Proving Ground (AVPG) Program. It will also
describe the need for new tools and data, including practical
flexibility in the NHTSA Part 555 exemption program to enable the near-
term development and long-term deployment of these safety-beneficial
highly automated vehicle (HAV) technologies.
The American Center for Mobility is a non-profit public/private
partnership. We are building a world-leading facility for innovation,
testing and product development, to act as a national proving ground
for future mobility. First and foremost, we are focused on public
safety, including the safe testing, validation, and self-certification
of connected and automated vehicle and infrastructure mobility
technology. In addition to testing, our mission is to accelerate the
development of standards, and to provide educational opportunities. ACM
incorporates a purpose-built test track environment directly integrated
with active on-road highway corridors, simulation laboratories, and a
combined corporate/academic technology park campus to facilitate
effective industry, government and academic collaboration and
information sharing, as well as a focus on educating the mobility
workforce for the future, STEM K-12 programs, and educating the public
in general.
It is clear we need a significant change in our transportation system.
Our past transportation system has served us well over the last 100
years, but is showing signs of strain. Every year we endure a national
tragedy of tens of thousands of Americans losing their lives on our
roads, in urban, suburban, and rural areas. That is the numerical
equivalent of over ten September 11 Attacks, or seven Iraq Wars, every
year. In many cities, we see ever-increasing congestion, with
accompanying loss of productive time, wasted energy, and unnecessary
greenhouse gas emissions. Largely we have accepted these undesirable
outcomes as the status quo. It is clear we need a significant change,
and we are now on the cusp of introducing connected automated vehicle
(CAV) technology that can begin that change.
Safety must be everyone's first focus. NHTSA's census of fatal car
crashes(1) shows that 35,092 people lost their lives in
2015. Historically, human error or actions account for, or contributed
to, 94 percent of these fatal crashes.(2) By reducing the
effects of human error, connected and automated vehicles have the
promising potential to reduce or even eliminate these fatal outcomes by
avoiding the crash scenario altogether.
Because of promising safety potential, the United States should focus
on developing and deploying CAV technology in the most
responsibly
expedient manner possible.
Transportation is the lifeblood of our economy and society. The
ability to efficiently and effectively move people and goods is
critical to the social and economic well-being of the United States,
and will help us remain competitive with other international economies.
The United States covers a very large geographic area, and part of our
industrial and economic strength is that we can transport people and
goods across that large area in a safe, efficient, and economical
manner. In addition to safety benefits, CAV technology provides the
opportunity for a ``systems approach'' to transportation, with
substantial potential improvements in mobility, energy use, equity in
transportation, and positive impact on the Nation's economy.
AV technology is being developed very rapidly, largely led by
industry, but aided by key frameworks, research, and support from
Government and Academia. While we have one hundred years of experience
with human-driven vehicles, we need to gather experience and data with
automated vehicles in a much shorter timeframe. Inherently, automated
vehicles will be data-rich due to the basic operational need to sense,
analyze, and act on data that is generated continuously through
operation. We need the initial experience of operating these vehicles
in sufficiently substantial numbers to generate the broad data across a
multitude of scenarios and environmental operating conditions necessary
to ensure safety, and to scale the technology to full deployment.
Over the last decades, automotive safety technology has progressed
significantly, and that progression has resulted in the savings of
hundreds of thousands of American lives.(3) The development
and introduction of each technology has required a very thorough
engineering process, including research, testing, product development,
verification, validation, standardization, certification, education,
and in-use monitoring. Even the simplest of technologies, the seat
belt, was conceived, designed, and introduced through this phased
process. Many safety technologies developed by industry and government
through this process have become mandated through regulation by Federal
Motor Vehicle Safety Standard (FMVSS).
Modern safety technologies have become more and more complex.
Airbags and Supplemental Restraint Systems (SRS), Automatic Braking
Systems (ABS), Electronic Stability Control (ESC), Forward Crash
Warning (FCW), Lane Departure Warning (LDW), Crash Imminent Braking
(CIB), and Vehicle-to-Vehicle Communication (V2V) have required ever-
greater developmental efforts, and are subject to factors outside the
control of the vehicle alone. However the basic engineering processes
listed above have proven capable of developing these systems to an
extremely robust level that is befitting of a life-saving technology.
AV technology will likely be the most complex automotive safety
innovation yet deployed, however the basic process honed over the last
decades will serve as a capable and required starting point.
Voluntary standards have proven to be a key step in the development
of all of these safety technologies. Industry and Government research
and testing generate the basic scientific and technical knowledge for a
new technology. This leads to standardization of key definitions,
designs, test procedures, validation models and methods, and
certification protocols that enable a new technology to be
commercialized. Often, these become codified in voluntary industry
standards, such as by SAE International and other Standards Development
Organizations (SDOs). Occasionally, but often for safety technologies,
these voluntary standards form the basis for Federal regulations as
codified by FMVSS, as shown in Figure 1.
It is critical to accelerate voluntary standards in order to ensure
safe
deployment of AVs.
AV technology is being developed and is advancing at a faster rate
than the traditional standards process can fully accommodate. SAE
international has begun to promulgate basic standards, such as taxonomy
and definitions, which have already served as the basis for the Federal
AV Guidance. Additional voluntary standards are needed immediately to
ensure that these new approaches in testing, validation, data
collection, data-sharing, privacy, cybersecurity, and others areas are
developed to ensure safety, while not inhibiting or stalling the
technology development. These standards will also likely form the
technical basis for future FMVSS requirements for vehicle performance
or equipment.
It is critical to accelerate these voluntary standards in order to
ensure safe deployment of AVs. From now through 2025, AV development
will continue to move very rapidly, and initial voluntary standards
must be in place by that time-frame to support the first significant
vehicle deployments. Other countries and regions, including China and
the European Union, are funding and working diligently on standards
efforts to support their deployments and industry partners. The
complexity of AV technology will require innovative thinking for
testing, simulation, validation, standardization, and certification
tools and methodologies to support these standards.
Voluntary standards must also be accelerated for purposes of
creating a collaborative, unified common approach, and avoiding a
patchwork of standards or regulations that could inhibit or stall the
technology development. Even a small number of differing or conflicting
standards or regulations would significantly inhibit the development of
AV technology. Consumers' interests would not be served if they could
not operate their vehicle, or a shared vehicle, across state lines.
Differing standards could certainly limit the safety, equity, and
economic benefit of automated vehicles, and would put the U.S. in a
compromised position compared to other countries and regions around the
world.
It is also important to consider that the pursuit of voluntary
standards does not preclude the promulgation of State or Federal
standards. Indeed it may be prudent to consider the adoption of a
Federal FMVSS framework standard that establishes key foundational
requirements, such as definitions, manufacturer identification and
reporting processes, data reporting requirements, exemption processes,
consumer notification and privacy, enforcement requirements, etc.
However it is too early to promulgate significantly detailed vehicle
performance or equipment standards, as the needed technical
requirements are not sufficiently developed, and currently there is no
agreement within the technical and stakeholder community on the nature
or specifics of such requirements. Additionally, the premature
promulgation of not-fully-developed safety standards could result in a
``false sense of safety,'' whereby manufacturers or consumers may
believe that the technology is more ready or capable than it actually
is, simply because it is advertised as meeting safety standards.
State regulations are also ultimately necessary for full scale AV
deployment. The traditional State roles dictating operational
requirements, such as insurance, registration, training and licensing,
driving enforcement, etc., are still required, and appropriate, for
AVs. States may also want to proscribe new regulations that require HAV
owners or fleet operators to make sure safety-critical vehicle control
recalls are completed in a timely manner. However, creating state-by-
state standards for vehicle performance or equipment could result in
this undesirable patchwork, and should be avoided for reasons described
above. Additionally, the creation of state-by-state vehicle performance
or equipment requirements may also contribute to the ``false sense of
safety'' discussed above.
USDOT's Automated Vehicle Proving Ground (AVPG) program provides key
infrastructure and framework for safe deployment of CAVs.
USDOT's Automated Vehicle Proving Ground (AVPG) program provides
key infrastructure and framework for the development of CAV products,
tools, and standards in a rapid and collaborative fashion. This will be
a critical asset to ensure safe deployment, as well as ensuring that
the U.S. remains competitive in AV industry and deployment. This AVPG
Program was established very recently, and has the mission to serve as
a Community of Practice (CoP) to enable USDOT, industry, and other
stakeholders to develop, test, and validate AV products, standards, and
supporting tools.
This CoP will convene and enable stakeholders from companies,
government agencies, academia, facilities, consumer groups, and other
organization, to share best practices and innovations for testing
operator safety, facility design concepts and details, facility
operational best practices and lessons learned, data acquisition and
analysis system concepts, and testing and analysis equipment best
practices and standards. This intellectual capital will form the basis
for voluntary standards and mandatory regulation.
AV testing and validation must occur through three coordinated
approaches: Closed-Track testing, On-Road testing, and Simulation, as
shown in Figure 2 below.
It is crucial that AV testing and development programs include
three key complementary and fully integrated tools: (1) Test-track and
laboratory tests that validate full vehicles, systems, and components,
under realistic, variable, repeatable, and controlled conditions; (2)
on-road tests that verify the systems' robustness in real life
situations, including some that can't be implemented, or even imagined,
in the above tools; and (3) detailed simulations, including roadway,
traffic, vehicle, sensors, drivers, infrastructure, etc. that
mathematically model the millions of technology and environmental
variations.
These testing tools and methodologies must be used in combination
in an iterative fashion. Simulation results help prioritize and define
the initial round of test-track tests. The test-track allows a safe,
controllable, and repeatable environment for development of
functionality in numerous scenarios, and provides sufficient confidence
to begin on-road testing, but can never accommodate all scenarios. On-
road testing provides identification and validation to real-world
behavior, including unanticipated, key scenarios, but often does not
allow repeatability, so key critical scenarios must be reproduced in
simulation and track testing. This process must be rigorously planned
and executed to validate artificial intelligence as a substitute for
decades of experience of a human driver. These new methodologies are
not yet fully defined, and are the subject of intense research and
rapid development.
USDOT's AVPG program contains a combination and range of track, on-
road, and simulation facilities and capabilities across wide geographic
areas and environmental conditions. This range of capabilities is
important for the development of these products and tools in a rapid
and collaborative fashion, and will be a critical asset to generate
experience and data to ensure the safety of AV products and that the
U.S. remains competitive in AV industry and deployment.
AVs present both challenges and opportunities regarding data and
data sharing, requiring that a large amount of data and information be
sensed, acquired, amalgamated, analyzed for rapid decision-making in a
wide variety of travel scenarios and environmental conditions, while
protecting consumer privacy and ensuring cybersecurity. This data must
then be acted upon through control decisions and operational
monitoring. This same data can enable understanding of what is going
right, or wrong, with an AV, and if shared, provides an opportunity for
many vehicles to learn from the experience of one. While data best
practices for AVs are still under significant development by vehicle
manufacturers, testers, suppliers, government agencies, and service
operators, generally, these data practices will likely include
established processes and tools for the collection of in-use event,
incident, and vehicle information data for crashes, malfunctions,
degradations, failures, and unintended operation outside established
operational domains, while maintaining and protecting consumer privacy
and manufacturer/tester confidentiality and security. It is expected
that this information will become extremely useful for vehicle
development and operations activities, especially in early years of
deployment, as well as accident/event reconstruction purposes.
The AVPG Program will be critical infrastructure for developing
concepts, structures, processes, tools, and programs to enable and
implement these data sharing activities that are required for safe
deployment, while maintaining consumer privacy and system
cybersecurity.
For these reasons, the AVPG Program overall will provide the
foundation and program for the development of safe testing, operation,
and deployment, including the necessary voluntary standards and
mandatory regulations. As described above, significant work will need
to be undertaken by expert technical and policy communities to
theorize, prioritize, draft, and codify these standards in a
coordinated yet accelerated manner. U.S. investment in AVPG
infrastructure, facilities, equipment, and programs is a vital
necessity to ensure safe deployment of AVs, and to keep the U.S.
competitive with other major auto-manufacturing countries who are
investing in similar facilities and infrastructure at a much more rapid
pace.
HAV Data collection will continue after initial deployment, and
Exemptions are important to allow that data generation.
HAV data collection and analysis must also extend through the
initial years or even decade(s) of deployment. In addition to data
gathered and analyzed throughout the product development and validation
process, it will be critical to gather data in the initial years of
product deployment, and perhaps even continuing ad infinitum. HAVs,
like human drivers, must be capable of operating in a wide variety of
travel scenarios and environmental conditions. No amount of testing and
simulation can guarantee that every possible scenario has been
encountered before initial deployment, just as human driver education
cannot accomplish that before a driver is licensed. Secondly, setting
of standards, especially government regulation is always aided by the
collection and analysis of real-world safety experience and data.
Because of this it is imperative to create, collect and analyze in-use
data for initial deployments of these HAV vehicles.
According to a 2016 NHTSA/Volpe study,(4) a number of
existing FMVSS contain requirements that either conflict with, or do
not allow, basic HAV design tenets, especially related to human
operated controls. In order to enable that early product deployment and
regulatory data gathering, and to enable HAVs in general for their
significant safety and mobility potential, it is important that
Congress and USDOT review the current authority, and identify and
implement a solution for exempting compliance of a sufficiently
substantial number of vehicles to FMVSS requirements that do not
pertain to HAV design or operation.
CFR 49 Part 555 currently allows NHTSA to temporarily exempt
petitioning manufacturers' vehicles from certain FMVSS requirements,
based on one or more of four situations:
(1) Substantial economic hardship
(2) Facilitation of development of new safety features
(3) Facilitation of development of new ``low-emission'' features or
vehicles
(4) Inability to sell a vehicle with an overall equivalent or
greater level of safety
It could be argued that initial deployment of HAVs would already be
covered by situations 2, 3, and 4. However the number of vehicles (up
to 2,500) and the time duration (up to 2 years) for the exemption is
somewhat limited, and may not serve the need of collecting a sufficient
body of data to ensure safe operation, and ultimately create
appropriate FMVSS requirements.
Notably, and importantly, situations 2, 3, and 4 include the
requirement that the petitioner demonstrate an equivalent or better
level of safety, as compared to a non-exempted fully conforming
vehicle, as part of the information for NHTSA's consideration.
As part of this recommended Congressional and USDOT review, it
should be debated whether situation 1 is applicable or advisable as an
exemption rationale for HAVs, especially since this basis does not
require the demonstration of an equivalent level of safety.
Regardless of the rationale for exemption, it remains imperative
for HAVs that the petitioner demonstrate, ideally through test-track,
on-road, and simulation data, that the exempted vehicle clearly
provides an equivalent or better level of safety, and that NHTSA have
the final judgment that that requirement is met. Lastly, NHTSA must be
authorized and adequately funded, including appropriate FTE levels, to
ensure that this exemption program can be enacted fully and
expeditiously so that it meets its full intent.
Closing
In closing, I wish to reiterate that we are entering a new era for
a greatly improved transportation system built on key technologies. The
United States must embrace and invest in these technologies to provide
increased safety, mobility, and equity for the travelling public, as
well as support the economic competitiveness of our automotive industry
and our country.
I urge Congress to consider the following:
Authorize and fund USDOT to direct, fund, and collaborate on
voluntary industry standards, including testing, validation,
cybersecurity, privacy, and self-certification tools and
methodologies, including appropriate FTE to implement this
program.
Authorize and fund the USDOT AVPG program to provide
facilities and opportunities for product development,
standardization, education, and public-private partnerships.
Consider revising NHTSA's exemption authority to enable HAV
deployment and early data collection, and authorize and fund
appropriate FTE to implement this program.
I appreciate this opportunity very much. Thank you for your
attention.
About The American Center for Mobility
The American Center for Mobility's proving ground is designated by
USDOT as part their AVPG Program. ACM's core mission is to enable the
safe development of automated vehicles as part of a future
transportation system. To perform that mission, ACM is committed to
develop and share safety-related approaches, information, and data in
both test facility design and test operations, as well as explore and
create data sharing opportunities for non-confidential, non-PI, testing
and operation data, as part of a similarly-committed Community of
Practice, while maintaining and protecting consumer privacy and
manufacturer/tester confidentiality and security.
With $110M in direct or incorporated investment, and leveraging an
additional $115M of regional assets, ACM is designed to be the premiere
national proving ground for shared-use mobility and advanced automotive
testing. Founded in early 2016, ACM is a joint initiative among the
State of Michigan, including the Michigan Department of Transportation
(MDOT) and Michigan Economic Development Corporation (MEDC), the
University of Michigan (U-M), Business Leaders for Michigan (BLM), and
Ann Arbor SPARK. ACM incorporates private funding in partnership with
public funding from the State of Michigan. ACM is strategically located
in southeast Michigan and is recognized as a centerpiece of the state's
``Planet M'' initiative representing the collective mobility efforts
across the state defining Michigan as the global center of mobility.
ACM's testing and validation capability is based on real-world
representative testing environments designed to address a wide range of
variations of pre-crash scenarios, including variable but controllable
critical near miss scenarios. This testing is critical to the proper
development and validation of CAV products and will serve AV
developers' needs for both hardware and software. ACM's testing assets
will also be critical to the work of government agencies and academic
researchers collaborating on safety, mobility, and energy aspects of a
CAV-based transportation system.
The American Center for Mobility program will focus on three
activity pillars that are critical to CAV safety: testing/validating,
standard setting, and education. CAV technology, including
communication, infrastructure, and cyber security, is developing very
rapidly. All three of these key ``pillar'' activities should be
advanced simultaneously to ensure that the technology is developed to
maximize safety, mobility, energy, transportation equity, and other
benefits.
ACM is working with other facilities, government agencies,
academia, industry, and like-minded experts, to share best practices
and innovations for testing operator safety, facility design concepts
and details, facility operational best practices and lessons learned,
data acquisition and analysis system concepts, and testing and analysis
equipment best practices and standards. ACM is convening operators and
designers of other test facilities to lead the effort in the formation
of this Community of Practice. ACM is partnering with standards
organizations, such as SAE International, IEEE, ISO, NIST, ASTM, and
others in these sharing discussions at the appropriate time, and will
work with those bodies to enable voluntary standards as appropriate.
ACM's mission includes the acceleration of these voluntary
standards through convening and the creation of laboratories designed
specifically for AV testing and standardization. ACM has partnerships
in place with SDOs which will enable an accelerated process that will
allow the standard setting experts to convene at ACM and participate in
testing and demonstrations that inform the strategy, details, and
ultimately decisions on the standards that are created. ACM provides a
unique venue to contemporaneously develop and validate standards in
rapid cycles and serve as a convening center where industry,
government, SDOs, and other interested stakeholders can collaborate and
accelerate the development of critical standards.
Through its ability and focus to convene technical experts,
numerous automotive development users, and standards bodies, ACM will
establish a user group to develop a uniform approach or voluntary
standard to data sharing. While it is unlikely that any manufacturer or
tester will share all of its vehicle and incident data, this user group
will act to establish a process to share noteworthy scenarios, and to
define a subset of data and data retrieval/access methods, that they
can agree to share. ACM will fully protect consumer and public privacy
and security, and will take steps to ensure that any data or
information sharing activities do not violate, hinder, or compromise
integrity of any consumer privacy/security agreements or arrangements
put in place by manufacturers, testers, agencies, public entities, or
by ACM itself. Similarly, ACM commits to maintaining the
confidentiality and security of proprietary confidential business
information and data on behalf of its users, testers, and private
partners.
ACM will lead the formation of a specific Community of Practice
focused on safe CAV testing, as a proposed subgroup of the
International Committee for Proving Ground Safety (ICPGS) established
to enable the rapid sharing of information, best practices, and data
including the following aspects: Best safety practices specific to CAV
testing, including vehicles, infrastructure and communication; driver
training (remote and in vehicle); safe conduct of testing and
operations; safety management plans and protocols; common safety
incident reporting protocol; and common data format and agreement to
share non-proprietary/non-confidential data.
ACM fully supports the establishment of a network of experts and
commits to providing engaged expertise through comments, meetings,
workshops, and more formal activities including volunteered
participation in an expert committee. In addition, the ACM facility
will serve a very important purpose to educate the current and next
generation of engineers, policy-makers, lawyers, and strategists and,
in league with a consortium of universities and community colleges,
will develop a curriculum to ensure that expertise is generated and
available for hiring or appointment throughout industry, government,
and academia.
References
(1) National Highway Traffic Safety Administration (2016, August).
DOT HS 812 318 2015 Motor Vehicle Crashes: Overview
(2) National Highway Traffic Safety Administration (2015,
February). DOT HS 812 115 Critical Reasons for Crashes Investigated in
the National Motor Vehicle Crash Causation Survey
(3) National Highway Traffic Safety Administration (2015, January).
DOT HS 812 069 Lives Saved by Vehicle Technologies and Associated
Federal Motor Vehicle Safety Standards 1960-2012
(4) USDOT John A. Volpe National Transportation Systems Center
(2016, March) DOT-VNTSC-OSTR-16-03 Review of Federal Motor Vehicle
Safety Standards for Automated Vehicles; Identifying potential barriers
and challenges for the certification of automated vehicles using
existing FMVSS--Preliminary Report
The Chairman. Thank you, Mr. Maddox.
Let me direct this question to the entire panel. In order
to realize the safety benefits of self-driving vehicles,
manufacturers and new market entrants must be able to test and
deploy on public roads and across multiple states.
Traditionally, as Mr. Bainwol pointed out, the Federal
Government has regulated the vehicle while states have
regulated the driver. While I believe that states have a role
to play in the deployment of these vehicles, we must be careful
to avoid a patchwork of regulations, as a number of you have
pointed out.
So the question is: Would you support a Federal framework
governing the safety of these vehicles? We'll start with Mr.
Bainwol and move across the panel.
Mr. Bainwol. Short answer, sure, absolutely.
Mr. Csongor. The answer would be yes, especially if it
simplifies our ability to deploy and test across the different
states. So, of course, the devil would be in the details, but
we test today in California, New Jersey, and we're able to
deploy and test vehicles, and there are certain standards that
are completely fine and very logical. I think the devil is in
the details, but I would agree with a framework of that nature.
Ms. Sheehey-Church. An even shorter answer would be,
obviously, yes, we do, but I'll broaden it just a little bit.
To avoid that patchwork, you know, of state requirements, it's
important to make sure that we balance the technology and the
regulation. So in saying that, it's important that NHTSA, the
auto industry, and safety advocates really get together and
make sure that the collaboration is key and ongoing, along with
the data that we will be taking and looking and watching. But
the short answer is absolutely.
Mr. Maddox. And, last, I would add a fourth yes to that
question, for sure. I think it's a necessary step, as I talked
about, to be able to create the data that we need to validate
these vehicles and, importantly, to educate the public about
what they can and can't do. I think a framework would be a very
good first step.
The Chairman. Many of the regulatory challenges for the
testing and deployment of self-driving vehicles stem from the
fact that current Federal motor vehicle safety standards, as
written, don't take into account self-driving vehicles. With an
increasing number of auto manufacturers and new market entrants
looking to test and deploy this technology in the next few
years, many have expressed the need for exemptions from certain
Federal standards to deploy vehicles on public roads.
I guess this would be perhaps more specifically for Mr.
Bainwol, Mr. Csongor, and Mr. Maddox. But why is the expansion
of current exemptions important, and how will that inform
future standards?
Mr. Bainwol. It's vitally important. We look at this as a
multipronged approach to maximize this transition. On the one
hand, you have to have Federal uniformity, but you also need
exemptions in a substantial number to generate--really, for two
reasons. One is to generate the test data so that we can get to
the point where we can deploy with confidence that's data
driven.
Second, there's a massive economic commitment here. This is
billions of dollars of R and D. In order to make it
economically attractive, to make the math work, you've got to
be able to deploy--not just test, but deploy, and it's going to
take time for the final rules to be written, and in this
interim period, you need a mechanism to deploy. So you need it
for two reasons, testing to prove a concept and, second, to
make sure that the math works on the capital investment.
The Chairman. Mr. Csongor?
Mr. Csongor. I think the expansion of the Federal
exemptions, particular to NVIDIA and to someone who is
developing the underlying computer technology that would make a
self-driving car possible--in my remarks, I commented on the
connection between getting to that level of safety and the
amount of data that's generated, and that we can ensure that
the data that's generated is as diverse as possible and that we
can get it as quickly as possible in order to deploy it.
I know that other issues will come up with regard to the
data, including cybersecurity and privacy. All of these are
important issues. But in terms of the Federal guidelines, from
a development point of view, access and development of that
data is critical to the safety of the car fleet.
The Chairman. Mr. Maddox?
Mr. Maddox. I would echo that, perhaps with a bit more
specific thought. Certainly, we need a significant amount of
in-use data to validate the technology and to give the vehicles
a chance to experience a given scenario and learn from it, and
then maybe even educate other vehicles or other manufacturers
how to handle that scenario. That's the way we learn as humans
to drive. We're taught the basics. We can't experience every
scenario before we're let loose, if you will.
But with automated vehicles, we need a significant amount
of data, and the number of vehicles to gather that data will
need to be substantial. So for that reason, I think raising the
numbers of vehicles that could be exempted would be a very
positive step.
The Chairman. Final question, Ms. Sheehey-Church. And,
again, I want to express my sympathy to you for your family's
loss and just appreciate you being here today to testify.
In your testimony, you highlighted the fact that,
tragically, the highest percentage of deaths that occur on U.S.
highways are due to drunk driving and are, therefore,
preventable. Can you elaborate on how you see self-driving cars
being part of the solution to drunk driving?
Ms. Sheehey-Church. Absolutely, Chairman. When we look at
the numbers, the 35,000, and you break it down, very simply,
10,000, as you said, for drunk driving. But the opportunity for
autonomous vehicles will allow for all those other human
behavioral errors and including drunk driving--again, MADD is
there--our mission is to eliminate drunk driving, and we'll
look and support any advanced technology.
So when we look at AV, it was something that we said--it's
an emerging technology that just has to be done, because it
will not only take care of the other behavioral errors that are
happening, you know, the drowsy or sleepy or seatbelts or
speeding, but it will allow also drunk driving. What we want to
do at MADD, we want to stop--we separate the drinking from the
driving. With AV, we're going to separate the driver from the
vehicle itself. And, again, we're looking at Level 4 and 5 when
I answer that question.
The Chairman. Thank you.
Senator Peters?
Senator Peters. Thank you, Mr. Chairman, and thank you to
all of our witnesses for excellent testimony and we appreciate
all four of you being here today.
This first question is for Mr. Maddox. Prior to becoming
the CEO of the American Center for Mobility, you developed and
implemented a connected and automated vehicle public-private
research and development partnership with the University of
Michigan. You and I are very familiar with the auto industry,
and we know that it is very competitive. Competition is healthy
as it leads to better products.
But we also know that in this new era of transportation
innovation, we're going to have to see an awful lot of
collaboration between the industry. We know part of these
technologies are connected vehicles that are going to be
talking to each other, and a General Motors vehicle needs to
talk to a Toyota on the road and to a Ford and a Nissan, et
cetera, et cetera. That's a whole new business model in terms
of some of the ways in which we use proving grounds and test
vehicles, and we need to have that collaboration to eventually
inform safety standards for development.
So my question to you is: How will the American Center for
Mobility facilitate this work? What are some of the partners?
How are you bringing industry together? How is it also
necessary for us to use facilities like the American Center for
Mobility to go much broader than the auto industry?
Mr. Maddox. I'll start with that last one first. Certainly,
the American Center for Mobility and the other nine designated
proving grounds around the United States--we are key
infrastructure. To prove these vehicles out, we need to do both
on-road driving, simulation, and certainly track testing, and
the proving grounds are the facilities to do that testing.
More importantly, as you point out, it's the facility--it's
a convening location for these companies and different
industries to come together and develop the technology
consistently and collaboratively. At the American Center for
Mobility, we certainly are working with auto makers and Tier 1
suppliers, but, in addition, communications companies,
technology companies, insurance companies. They all have a
stake in understanding how the products work, how they don't
work, and to develop them so that they actually work together
for the public.
Importantly, we believe, all of those industries as well as
government has a stake in educating the public about what the
technology does or does not do, how it functions, what its
implications are. So these proving grounds also will serve as a
locale for educating the future workforce, the public, in
general, as well as each other.
Senator Peters. It's obviously very important to be doing
all of that, and it's important for us to move the development
forward, but also to understand that we're in a global economy.
We're going to have significant global competitors. A number of
other countries are also racing to get to this technology, and
whoever gets there first or is very early to get there will
have a significant competitive advantage.
Tell us a little bit about these centers, like the American
Center for Mobility. Where are other countries, who are our
competitors, and do we have to really get moving even quicker?
Mr. Maddox. That is absolutely right. Other countries are
moving very quickly. This technology was largely invented in
the United States, and I think we're still in the lead on a
technical level. However, when you look at the funding and the
activities and the programs that other countries--China, Korea,
Japan, the European Union--China alone has built two facilities
already and has plans for five more for a total of seven.
They are, like us, focusing on technology development and
standardization, because they know that the countries and the
regions that set those standards will certainly create an
opportunity for their industries that other countries don't
necessarily have. So we see that activity. We see them coming
to Ypsilanti, coming to Michigan, coming to other facilities
around the United States on a regular basis. We know it's going
on.
Senator Peters. Thank you.
Mr. Csongor, I'd like to have you elaborate a little bit on
some of the comments you made in your testimony related to
artificial intelligence, deep learning, machine learning, and
why really moving this technology forward in the auto industry
is critical, not just for autos, not just for the incredible
advances in safety, which we've already articulated, but for
the entire economy and for every industry. It has been
described to me that this auto technology is the moon shot for
artificial intelligence, and when that happens, we'll be able
to have incredible advances in every industry imaginable.
Is that accurate? Where do you see this going? Why is it
important to get it right with autos?
Mr. Csongor. I think that's very accurate, Senator Peters.
I don't think it's an over-exaggeration or hyperbole to say
that AI is the fourth industrial revolution, and what that
means is that everything will change, and it's specifically
targeted, as any industrial revolution, against problems that
exist within society. Many of the problems that we have can be
broken down to computational problems. The pursuit of a cure
for cancer is a computational problem.
We have enormous amounts of data being generated in every
field across so many industries. We just can't make sense of
the data. What's really required and the real--I think the
really critical role of AI is to get insight from large amounts
of data. So if you gather large amounts of data on speech, you
can understand idioms of human speech. You can create automatic
translation literally out of a Star Trek episode, where you're
able to speak one language and you hear another language out
the other side.
You can collect information on the oceans and have an AI
agent that sifts through the data to look for important trends.
And, of course, you can break down the world--highways, roads--
and then sift through that data, and then be able to identify
all the important things that matter to driving a self-driving
car.
Just imagine trying to break down the world into objects,
all the objects that we see when we drive a car. Imagine that
you had to have a person write manually software code to
describe what that object is. This is a stop sign. This is a
stop sign at an angle. This is a stop sign at night. This is a
stop sign that's covered with snow. We're able as humans to
recognize objects even though they're hidden or what we call
occluded, and deep neural networks are very good at that kind
of thing.
So I think that as we as a society evolve into a world
where we have lots of data, I think the ability of AI to
provide insight in that data is going to provide breakthroughs
everywhere across every industry.
Senator Peters. Thank you.
The Chairman. Thank you, Senator Peters.
Senator Moran?
STATEMENT OF HON. JERRY MORAN,
U.S. SENATOR FROM KANSAS
Senator Moran. Chairman, thank you and Senator Peters not
only for the hearing but for your efforts to determine
guidance.
The panel gets your blood flowing, excited about the
potential of what this means, not just for driverless vehicles,
but as you're describing what it means for us as a society, a
country, and a world. I'm having trouble putting my hands
around what it means in rural America. Incidentally, I come
from a place in which driving is still enjoyable. We look for
it. There's still occasionally Sunday drives, and then you add
to that farm equipment, and it is a different environment.
So, first, I would start by asking do we see this as
something that is, at least for the foreseeable future, limited
to cities and interstate travel?
Mr. Csongor. Maybe I can take a quick shot. There are two
comments I'd like to make. First of all, the use of AI in a
car, in our vision at NVIDIA, can really accomplish two things.
The first thing is the AI can drive the car for you if you
want. We all refer to a self-driving car as one thing, but in
actuality, this technology will likely be deployed in small
steps toward a big vision.
The big vision is a self-driving car that can drive very
largely autonomously, but I think the idea of a self-driving
car that can drive anywhere, absolutely anywhere, is not true.
If you drop a self-driving car in the middle of Greenland, it
will not drive on its own, just like a human would not be able
to drive on their own.
So one part, I think, is you already have solutions on the
road today, and then you're going to see every year products
come out that are better and better and better, and they'll
have more and more features and more and more capability.
However, in addition to self-driving, we actually believe that
the AI can be a co-pilot, or think of it as a guardian angel.
So with a camera inside the car that's connected to the cameras
outside the car, for example, we can track the position of the
driver's head, their gaze, if they happen to be looking to the
side and a pedestrian steps off on the right, but you're
driving, the AI can say, ``Attention, there's a pedestrian on
the right.''
So the use of this technology, I think, is not just limited
to self-driving but also to assisting the driver and being
there in case you'd like to drive because you enjoy driving.
The second comment is regarding your question on rural
areas. The system that I described to you on the self-driving
does not require that the car be continuously in connection
with a central computer. So in the system that I described, the
algorithms are tested with massive amounts of data, and then
they're over-the-aired to your PC.
So imagine your computer that you can get an update on, and
once you have the update, you can go anywhere with the
computer, and then at some point, you get back to an area where
you have a connection, and then as soon as you have a
connection, if there's an update, then you can do it. So I
think the architecture of the self-driving car that we're all
working toward does not require you to be in continual
connection.
Senator Moran. Thank you.
Mr. Bainwol?
Mr. Bainwol. If I can add perhaps a less sophisticated
answer, if you're in Russell, Kansas, you may not have the same
level of ride sharing ultimately. But you certainly will
benefit from the access questions. You know, if you're blind or
disabled in Russell, Kansas, it's the same benefit to be able
to get a self-driving vehicle to take you to your destination.
So that's part one.
Part two is, ultimately, this is all about crash avoidance,
and it doesn't matter where you are, crash avoidance inures to
the benefit of everybody. So you're avoiding accidents, you've
got the property damage issues, you've got the time issues, you
have the fuel savings questions, but, most of all, you have the
saving of life.
So I would argue that there are certain applications that
may be more prevalent in urban centers. But as a bottom line,
this is good for rural America just as much as it is for urban
America.
Senator Moran. Thank you very much. Let me see if I can one
more question. The question is related to insurance. I chair
the Subcommittee that has jurisdiction over insurance in this
Committee.
Let me ask you, Mr. Bainwol. The question really is: what's
the private sector role in insurance? Volvo stands out to me.
In 2015, they announced that the company would accept full
liability whenever one of its cars is in autonomous mode.
What's the industry's expectations? And then, second, what's
the request of Congress? What's required here in regard to
insurance?
Mr. Bainwol. Well, the insurance piece has a couple of
different dimensions. There's the question of liability, and
certainly on the question of aftermarket, the OEMs would like
to be protected from aftermarket adjustments to the vehicle. I
think there is a range of perspectives on liability on the
system itself from the OEMs, and that's something that we're
just going to have to work through. There obviously is a tort
system in place as well as broad enforcement authority, and
that does govern behavior in terms of the introduction of
technologies and the risk taking.
The other point on insurance that we often hear about is
that premiums will go down, and, ultimately, that may, some
say, damage the insurance industry, but at the end of the day,
if you have fewer accidents, that's a good thing, and claims
will go down.
When Volvo introduced the XC60 originally in the U.S. five
or six years ago, it was the only vehicle that had automatic
braking on the sport utility. No others did, and all of the
sport utilities that Volvo introduced did. So you had a nice
control group, and it was found that the incidence of claims
for the first six months was 27 percent down. So automation has
really a fantastic opportunity to change the very essence of
the relationship, and the need for insurance will go down. But,
ultimately, the payout on insurance will also go down.
Senator Moran. Thank you all very much. It's nice to get
excited about something as compared to worrying about
something. This gets me excited. Thank you.
The Chairman. I think this probably gets your staff and
your family more excited than you, Senator Moran.
[Laughter.]
The Chairman. He may not be out there driving on his own.
Senator Cortez Masto is up next.
STATEMENT OF HON. CATHERINE CORTEZ MASTO,
U.S. SENATOR FROM NEVADA
Senator Cortez Masto. Thank you, and thank you for this
incredible panel and discussion.
Two weeks ago, I was back in Nevada, and I had the
opportunity to walk through the Tesla site and plant that we
have there and drive in one of their autonomous vehicles. The
technology was absolutely incredible. I am very, very excited
about this phase.
One of the interesting aspects that I've seen in
transportation, particularly in Nevada, is the concept of smart
communities. I want to follow up on what you're saying, because
what we were talking about and what I heard was that these
autonomous vehicles are just one part of a larger community--of
smart communities, the Internet of Things and the connection of
things. I'm actually working on legislation to try and continue
that momentum for our communities to invest in smart
communities.
I wonder if anyone could elaborate on how you see this
fitting into that larger goal of smart communities that I know
so many communities are looking at right now.
Mr. Maddox. I'll start, if that's okay. Yes, we believe
that smart vehicles, automated vehicles, connected vehicles are
an important piece of a smart community, and really, a smart
community is all about data, how you collect data, and,
importantly, how you use data, and that is a very difficult
challenge. Certainly, automated vehicles and the way that we'll
move people, the way that we'll move goods will be a huge
contribution of that data source to that overview of how a
smart city is operating. So we know that these automated
vehicles will be that important piece.
The question is: how do you integrate all this data
together? And, frankly, we don't know that anyone has got that
real answer yet. That's one of the key reasons that we need to
do these early deployments so that we can actually get
experience with the vehicles in that smart city setting and
figure out how to use that data.
Mr. Csongor. First, I just want to comment on your ride in
the Tesla. I actually own one myself, and I share your
enthusiasm. It is not a new feature like a seat warmer or a
premium sound system. Traffic has just such a terrible effect
on your life that the absence of having to drive in it just
improves the quality of your life. So I do believe that beyond
safety, it's just delightful to have.
With regard to smart cities, NVIDIA has actually invested
very heavily in extending our technology into smart cities. In
many of the applications that you look at, whether it's
consumer applications from Facebook or Google or whether it's
self-driving cars or smart cities, there are common elements of
technology. For example, all of these feature and require the
ability to do image recognition, intelligent analysis of video.
So, for example, there's a concept called IVA, intelligent
video analysis. So being able to sift through lots of data to
be able to understand, either for security purposes or for
crowd management or, you know, things like this, a lot of these
same algorithms can be applied into smart city types of things.
Beyond that, there are other communication methods, vehicle
to vehicle, vehicle to infrastructure, where you can have cars
communicating with traffic lights, and, in theory, in theory,
the logical extension of all of these would be a society where
you don't have traffic lights or stop signs, that you're able
to have traffic flow completely naturally, intersecting with
each other. So there's quite an enormous amount of potential in
combining AI elements in the vehicle as well as the city,
because the infrastructure would have to expand to support a
world where you had autonomous vehicles and you can realize a
lot more benefits.
Senator Cortez Masto. Thank you. I know my time is running
out here.
Very quickly, first of all, Ms. Church, thank you for what
you do. I had the opportunity to work very closely with not
only MADD but Stop DUI in Nevada, and thank you for that
collaboration, because it wasn't just drunk driving. It was
distracted driving, it was driving under the influence that we
worked very hard to eradicate, and I think this technology will
help us address this issue, working closely with law
enforcement.
I think the data, besides insurance, will also help law
enforcement on so many levels that we can't even contemplate
right now. But one of them I do have a concern about is
cybersecurity space, this space of how we address hacking into
this technology. Do you think that, right now, we are
adequately staffing, and are we already preparing to address
cybersecurity needs as we build this infrastructure out? I'll
open it up for anyone.
Mr. Bainwol. So cyber is a real concern. It's one that
everyone shares. The fundamental proposition here is big data
provides an opportunity for great benefit, but we've got to
find a way to maximize the benefit and minimize the challenges,
and cyber falls into that category.
Several years ago, the OEMs came together and formed an
ISAC as other sectors have. What was unique was that we did it
in advance of an event, and to date, there has still not been a
market event. There has been wide hacking, but not a malevolent
attack. We've been fortunate--knock on wood. But as we formed
the ISAC, we promulgated best practices. The product cycle of
vehicles starts with security by design.
So we can always do more. It's not a static moment. This is
a journey, and the challenge will grow over time. But there's a
commitment to meet that challenge.
Senator Cortez Masto. Thank you.
Mr. Csongor. Of course, the subject of cybersecurity is
extremely important, and I know that there has been publicized
instances of vehicles that were hacked; however, I do think
it's important to note that the automobile is in the middle of
transformation. It was not very long ago when the most
sophisticated computer in that car was the radio. The
automotive is now becoming a computer, and in that recognition,
that's a challenge, and it's something that we have to do.
But the good news is this is not the world's first
computer. The computer industry has been building computers for
50 or 60 years, and they're used largely in autonomous vehicles
today. The aircraft that we fly today are powered by computers.
They're secure. There's computers powering every facet of our
society and our lives.
So I believe that there is a lot of knowledge on
cybersecurity. It just needs to be applied to automotive.
NVIDIA has been developing solutions for 20 years. If we did
not provide secure functionality in our computers, we would not
be in business. So I think this is a matter of application.
Mr. Bainwol. And we have about 18 suppliers in our ISAC,
and we welcome you joining the party.
Senator Cortez Masto. Fantastic.
Mr. Maddox. One last comment, if I could, on cybersecurity.
If we think about cybersecurity, we should consider security of
the entire transportation system. Certainly, the vehicle is
part of that, but so is the infrastructure, the traffic
controls, signs, and growing increasingly, even our personal
mobile devices will be part of that mobility infrastructure. So
when we think about cyber, we should consider cyber for all of
that.
Senator Cortez Masto. Great. Thank you so much. Oh, please.
Ms. Sheehey-Church. I'm obviously not the engineer on this
panel, but I know that MADD is looking at this issue and knows
that the auto industry itself is taking it very seriously. We
look at maximizing societal benefits while minimizing the
disbeliefs about it or minimizing the benefits about it, but
the fact is that the Auto ISAC is tracking and sharing and is
accountable of all the intelligence that we started in January
2016.
So the possibility is obviously always there, but they are
really overcoming the threats with the help of our automakers
and their suppliers. So I think they will have a good track
record going forward.
Senator Cortez Masto. Thank you. Thank you all.
Thank you, Mr. Chairman.
The Chairman. Thank you, Senator Cortez Masto.
Senator Gardner?
STATEMENT OF HON. CORY GARDNER,
U.S. SENATOR FROM COLORADO
Senator Gardner. Thank you, Mr. Chairman, and thanks to all
of you for being here today.
I was going to share my Sunday drive routine with Senator
Moran in the 1971 International Harvester Travelall that I
finally got running. So the only autonomous thing about that is
its consistent breakdowns.
Mr. Bainwol, I'll direct this question to you. Colorado has
been one of the states at the forefront of autonomous vehicle
technology. The Colorado Department of Transportation has been
pursuing rolling out their new RoadX program. It uses
innovative transportation systems to improve traffic flow and
modernizes the infrastructure itself to allow autonomous
vehicles to move safely and effectively.
It's incredibly important as you drive up to the mountains
from the airport when you fly in to ski. There's really only
one way that you can get there quickly, and it's a little bit
of a chokepoint at times. In October of last year, a fully
autonomous truck made a commercial shipment from Fort Collins,
Colorado, to Colorado Springs, and because it was Colorado, of
course, it was a shipment of--no, not that. It was beer.
[Laughter.]
Senator Gardner. With Colorado as one of the fastest-
growing states in the nation, I'm excited about the
opportunities created by autonomous vehicles to reduce
congestion and prevent accidents. In your testimony, you talked
about timelines for Tiers 3, 4, and 5 highly autonomous
vehicles. You talked about Tier 5 and that it can operate
seamlessly anywhere, and a person can drive--you know, anywhere
a person can drive a conventional vehicle, and that you think
it would hit the roads around 2025. But going on with your
testimony, you talk about how you don't think these vehicles
will make up a majority of the cars on the road for three
decades or so.
So what are the first sort of changes that we will see in
things like urban planning as autonomous vehicles hit the
market?
Mr. Bainwol. You're asking the tricky but correct question,
because we're going to have a situation of fleet mix for
decades. The estimates and timing were actually Moody's, not
ours, but it's all predicated on the reality that the average
age of a car is 11 years old. So if you go down to the 20th
percentile, they're really old cars, and so once you introduce
the technology, it takes a long time to wind its way through
the fleet.
So planners have to deal with a transition that's going to
occur for 20 or 30 years, and there's no magic answer. Part of
it depends on how aggressive a city or a state approaches it.
Part of it is not knowing exactly what the adoption is going to
be. There are fundamental questions. We don't know whether VMT
is going to go up or go down. We don't know whether car
ownership is going to decline and by how much versus ride
sharing.
So there are lots of open issues here, and you almost have
to place a bet and then try to mobilize and accelerate the
future. There are lots of crazy things. I met with a very
provocative nuclear physicist who talks about the future all
the time, and his comment, unrelated to his profession, was
that at some point in the future, metros will become basically
parking stations, because the last mile value of autonomy will
be so cheap that metro systems will end up collapsing, so
they'll become storage for something else. What do you do about
parking, and how long will that transition take?
So I don't really have a great answer. But the question is
profoundly on target, and planners are going to have to wrestle
with this question in a really profound way.
Senator Gardner. I'm happy if anybody else wants to take a
shot at that. Now, the technology you were talking about--is it
Csongor? Is that how you say it?
Mr. Csongor. It's pronounced Csongor.
Senator Gardner. Will AI be able to pronounce that?
[Laughter.]
Mr. Csongor. Probably not.
Senator Gardner. All right. Please go ahead if you have
anything else.
Mr. Csongor. I think one of the interesting next ripple
effects as a result of AV, autonomous vehicles, will be a big
problem in our cities and our infrastructure, which is parking
lots. Today, a lot of urban sprawl is caused by the fact that
every building has to have a parking lot right beside it,
because the people driving don't want to walk for a long period
of time.
NVIDIA is actually--we're building a new corporate building
in Santa Clara, and if you look at it, there's no parking lot
on the outside. Now, the parking lot will be underground, but
the AV capability introduces the possibility of being able to
free up a lot of land, to eliminate a lot of this sprawl, and
then to simply have an autonomous vehicle go park and it
doesn't matter to you how far away it is, as long as it's
somewhere close by. So I think that's an exciting----
Senator Gardner. Well, I look forward to carrying on the
conversation about the rollout of Tier 5 and other technologies
and the integration of Federal, state, and local regulations
and policies as we move forward.
Really, Mr. Bainwol, to your point of whether a metro or
something like that is necessary as we're pursuing
appropriations to put billions of dollars into certain public
infrastructure projects, at what point does the intersection of
cost to benefit, because of AV technology, start impacting
those decisions? So that's a question we're going to have to
answer.
So thanks very much for your time.
The Chairman. Thank you, Senator Gardner.
Senator Young.
STATEMENT OF HON. TODD YOUNG,
U.S. SENATOR FROM INDIANA
Senator Young. Well, thank you, Chairman, for holding this
hearing, and I want to thank our panelists for their testimony
here today. I really appreciate it.
Today, private industry is taking the lead in developing
many of the standards that are driving how we talk about
automated vehicles. For example, the Society of Automotive
Engineers has issued standards defining six levels of
automation. Many governmental bodies, including our Department
of Transportation in the Federal automated vehicles policy,
have adopted or referenced those standards.
That strikes me, generally speaking, as a good model to
continue as technology is still developing. That said, if we're
to preclude states from enacting safety standards for fear of a
patchwork of regulations, something that's been invoked here
today, we may need to have some minimum level of reporting to
NHTSA.
Are you concerned--anyone can speak to this--that new
entrants and startups in the industry might have difficulty
working with NHTSA?
Mr. Bainwol. Concern may not be the right word. But I think
everybody who wants to enter the space ought to. The rule set
should be technology neutral, and at the end of the day, for
common opportunity, there should be a common obligation to meet
the safety standards. The entire enterprise is predicated on
the notion that we're going to improve safety, NHTSA's reality
is a positive reality. We've had 50 years of safety standards,
and the CDC describes the outcome as one of the great public
health achievements of the century. So we want to recognize the
value that that system provides and make sure that everybody
complies.
Senator Young. So we see this across industry sectors,
right, where the incumbents oftentimes play an important and
constructive role in developing new standards, but sometimes
those standards are also tailored to advantage their business
models, right? They have to answer to shareholders. They're
publicly traded companies.
So to ensure as much dynamism, as much innovation, so we
can best serve the consumer markets, which all of us want, are
there specific steps that anyone has in mind that can be taken
to ensure that we're responsive to these new startups?
Mr. Maddox. I'll add to that, if I could. Certainly,
voluntary standards are a critical step. I discussed that a
little bit earlier. The key thing we need to ensure is that
those voluntary standards efforts are really directed and
focused on an outcome that can result in a FMVSS and result in
the ability of manufacturers, be they new or existing, to
deploy their vehicles so that we can learn from them.
So I think the critical piece is that U.S. DOT, NHTSA,
perhaps directly, should work together with SAE and ITE for
traffic engineering and IEEE for communications to make sure
that those voluntary standards efforts are directed to be
something that could be useful for a future FMVSS standard.
Senator Young. Mr. Csongor? If I said that correctly?
Mr. Csongor. It's OK. The Cs are like Ch.
Senator Young. All right. Csongor. All right, sir.
Mr. Csongor. It's OK. Startups are an important part of any
technological disruption. They're a hotbed of innovation, a
hotbed of ideas. When I joined NVIDIA in 1995, we were one of
them, and because of the climate that was made possible by our
government and by Silicon Valley, we were able to grow and make
a difference in the world. So our DNA is wired to not just
permit, but encourage the development of breakthrough
technology on top of our platform.
When we work with startups, they are all aware of standards
from committees or industries, ASIL ratings, and they are able
to work with them. I think as long as regulation creates
reasonable compliance rules for reporting information and
things like this, I think that we're going to have no problem
and we're going to enjoy a lot of benefits from the startups.
Senator Young. Thank you.
Mr. Maddox, as I conclude here, you know, you might imply
or infer from my previous line of questioning that I'm
concerned about Congress or Federal agencies or, frankly, some
incumbents putting their thumb on the scale to the disadvantage
of others. SAE is obviously at the forefront of developing many
of the voluntary standards you've discussed. Are there other
standards development organizations that should be part of the
conversation as we move forward, domestically or
internationally?
Mr. Maddox. Certainly, if you think of a future
transportation system as a system, which includes vehicles,
roads, communication systems, then there are other standards
organizations like Institute for Transportation Engineers, ITE,
based here in the U.S.; IEEE for communications. Those will be
critical to bring together, and now we're talking about a very
difficult lift, but an important one, which is getting
standards bodies to talk to one another, just like the vehicle
manufacturers and the other technologies have to. So those are
critical stakeholders in the standards game.
Certainly, there is an international standards effort, and
we always have a goal to try to harmonize vehicle standards and
other standards internationally. It's very difficult to do
that, but the best time, the best opportunity to do that is
when the technology is nascent. So we should involve those
other international bodies also.
Senator Young. Well, thank you. It sounds like we need some
standards for the various standards organizations discussing
these topics together.
Thank you.
The Chairman. Thank you, Senator Young.
Senator Inhofe is up next.
STATEMENT OF HON. JIM INHOFE,
U.S. SENATOR FROM OKLAHOMA
Senator Inhofe. Thank you, Mr. Chairman.
So that a lot of the people here and at the table be aware
of it, one of the problems we have is we have lots of meetings
happening at the same time, and it happens that half the
members of this committee are also on Environment and Public
Works, and we're meeting at the same time. So if I ask
something that has already been asked, you can just be
offended.
[Laughter.]
Senator Inhofe. Mr. Csongor, to facilitate the deployment
of autonomous vehicles, there's a lot of conversation about the
types of Federal regulations that should be in place. One thing
I'm concerned about that DOT is guilty of is setting up a kind
of system where new technology cannot be quickly adopted
because of an onerous approval system. This is exactly what
happened with FAA. When we get some things as innocuous as
applying GPS into an iPad system that you can use, it actually
took years to get that done. So I am concerned about this,
because, you know, that's government.
Does that concern you, too? And tell me why if it does not.
What can we do now to put something in place to keep up with
this?
Mr. Csongor. I think that the path that we're going down,
which is--and the spirit of this Committee, I think, is not
just to ensure the safety, but you can feel that the spirit of
this committee is to make sure that development work goes on,
whether it's from infrastructure or startups or all the people
here on this panel, and that we can streamline the effort to do
it.
So I feel that we all understand the goal. We understand
the benefits of what's being--what's at stake, and there might
be just a few small steps, you know, to focus on first, and I
had mentioned it earlier in my comments, but I'll cover it
again. The nature of the technology of an autonomous vehicle
through AI is one where access to large amounts of data that's
very diverse is very important. So it has been mentioned a
number of times, but, certainly, unified regulations and then
enforcement through NHTSA to provide that single unified
compliance would dramatically help us in developing cars.
Senator Inhofe. I'm glad to hear you say that your
observation is that this Committee has a reputation of getting
things done and avoiding some of the unnecessary steps that
other bureaucracies do. So I think this committee should be
complimented, and I think you are accurate when you say that.
Now, when you talk about all the data that's going to be
necessary, Oklahoma is proud to be a data center state, and
Google has a massive facility that we're very proud of. It's in
Pryor, Oklahoma. It's supplied with power from the GRDA, Grand
River Dam Authority, in Oklahoma, the entity that generates
power off of a large lake.
Autonomous vehicles will be collecting all kinds of data,
pictures, and video every second that they're in operation, and
this is going to require vast amounts of data stored at
facilities like Google's. But what we don't want is the wrong
people to get ahold of that.
Have you discussed in some detail safeguards that can be
made? Because that threat is out there. There will be a lot of
the wrong people wanting access to stuff that we don't want
them to have access to. What are your thoughts about that?
Mr. Csongor. I agree with you very much. I can speak on
behalf of NVIDIA, and I can talk to some of the steps that we
take or the protocols that we put in place. First of all, the
data that we collect, we use for one purpose and one purpose
only, and that's to improve AI algorithms so that you can have
a safer car. We regard any of the information that is collected
as a part of that data as private information. We don't share
it with anybody unless we have a very secure way of sharing
that data.
The data collected in our vehicles, if it's updated in a
system where you have an over-the-air update, the transmission
of the data, the upload and the download, is encrypted. And
then within the building, there's a multilayer security
protocol, ranging from the building security and the server
infrastructure, and then within the server, there are various
layers of firewall. So I won't get into all the details. I
think that, again, the data that we collect, we treat as if
it's the data that is our secrets, the secrets of our use and
everything, and then we protect it as such.
Senator Inhofe. Sure. Well, you don't need to detail any
more of it, but it sounds like you're on to this and you
recognize that the problem exists, and you're in a position to
take care of it.
Thank you, Mr. Chairman.
The Chairman. Thank you, Senator Inhofe.
Senator Hassan?
STATEMENT OF HON. MAGGIE HASSAN,
U.S. SENATOR FROM NEW HAMPSHIRE
Senator Hassan. Thank you, Mr. Chair and Ranking Member.
To the panel, thank you all for being here this morning.
When many individuals discuss the development and scaling
of automatic vehicles, it's often in the context of breaking
down barriers for individuals who are not currently able to
drive a vehicle. In 2015, the National Council on Disabilities
released a report titled ``Self-driving Cars: Mapping Access to
a Technology Revolution.'' This report recognized the promise
that this technology holds for individuals with disabilities
while at the same time highlighted some challenges of
independent use of these vehicles, such as the integration of
accessibility and assistive technology, mapping system
accuracy, and equipment failure.
Keeping these challenges in mind, I'd ask anybody on the
panel who would like to address it: What role do you see in
automated vehicles providing greater access for individuals who
experience disabilities? And then the add-on to that is that
we've seen time and again that advanced technology best serves
individuals who need it most when those individuals are part of
the development process to ensure accessibility. So what can
the industry and local, state, and Federal Government do to
ensure that people with disabilities are included in the
development and rollout of this technology, from licensing
issues, insurance, design, usability?
Mr. Maddox. I'll start.
Senator Hassan. Sure.
Mr. Maddox. I wholeheartedly agree that automated vehicle
technology could be extremely beneficial for those who don't
have access to transportation for many different reasons.
Equity in transportation is something that I think, over the
next 20 years, we will build a greater and greater, more
thorough understanding of and a more comprehensive solution set
than we have today. There's no question about it. Automated
vehicles will play a strong role in that.
I think it's critical if we think about how to accelerate
automated vehicles--and this is really Level 4, Level 5, the
highly automated vehicles. Demonstration programs really are
key. We need to actually put them on our streets, first in
small numbers, but in ever-increasing numbers. That will
provide the opportunity for this group of folks who don't have
equitable access now to really experience the technology, and
it does come back to consumer education. It is critical to
educate every transportation user, and that's probably
everybody in this room, for sure, if we walk or drive to work
or to the office.
So I think that really involving that community--the best
way to do it is through demonstration programs where they can
learn firsthand and provide input to the designer of the
vehicle or the system.
Senator Hassan. Thank you.
Mr. Csongor?
Mr. Csongor. This is a subject particularly dear to my
heart. My mother lives in a retirement community, and I've had
a chance to talk to them, and they're very excited about this
potential. I think that one of the interesting applications of
AV vehicle technology is actually even beyond the self-driving
cars that he mentioned, but also in the introduction of
autonomous scooters or shuttles.
We're working with startup companies in shuttles and
scooters that actually have the potential to be deployed much
sooner to these people, and the reason being that, for example,
if you had shuttles within a community, a retirement community,
it's what we call a geo-fenced area. You don't have to have it
operate all over the world. So the potential for rapid
deployment to help these types of communities is actually very
exciting.
Senator Hassan. Anybody else want to comment? Mr. Bainwol?
Mr. Bainwol. I'd just add that's precisely the right point.
In a geo-fence context, it's going to happen sooner rather than
later. To realize the full potential of what you're asking
really is Level 5, though, and as we've talked about, the
transition is a long one. So from a congressional perspective,
if that's the focus, then the mission is to accelerate that day
that we get to Level 5, and that means public policy that is
geared toward accelerating the future.
Senator Hassan. I would just ask you all--as we think about
piloting, getting upfront input from people who experience
disabilities before the product is finalized is really, really
critical. We see this right now. We're having to upgrade and
change a lot of existing technology for accessibility purposes,
and if you can include the community sooner rather than later,
I think it streamlines the whole process, and you learn that
when you address issues with people with disabilities, you
often address convenience and other issues for the general
public as well.
Because I am running out of time, I will make sure we
submit this question to you all in writing. But, you know, the
other thing that strikes me is as we discuss infrastructure,
generally, for transportation in this country, it would be very
useful to get your input about what a new generation of
infrastructure should include, whether it has particular kinds
of street lights, cameras, clear roadway lines, adequate
signage, et cetera, to accommodate these vehicles. And I'll
look forward to your written answers to that. Thank you very
much.
Thank you, Mr. Chairman.
The Chairman. Thank you, Senator Hassan.
Senator Blumenthal.
STATEMENT OF HON. RICHARD BLUMENTHAL,
U.S. SENATOR FROM CONNECTICUT
Senator Blumenthal. Thanks, Mr. Chairman. Thank you for
having this hearing.
Ms. Sheehey-Church, I want to thank you for coming from
Connecticut, from Madison, where I believe you live, and for
sharing your really tragic story about your son. As a father of
four, I can't imagine your loss. It's every parent's worst
nightmare. So I'm particularly grateful that you've come to
Washington, but also for your tremendous work and that of
Mothers Against Drunk Driving throughout Connecticut and our
Nation.
I've been very, very privileged and proud to work with MADD
over many years as Attorney General and now as a U.S. Senator
and know full well that too many families have been devastated
by this continuing scourge. We tend to focus on the issue of
the moment while young people, really people of all ages,
continue to perish as a result of this fully preventable
condition.
I would like to ask you perhaps if you could reflect why we
should be especially careful about semi-autonomous vehicles and
the possibility that it may create a false sense of security?
Ms. Sheehey-Church. Thank you, Senator Blumenthal. I think
in the beginning when I talked about MADD's concerns and MADD's
four principles, we were really looking at 4 and 5. That's the
level that we have to look at, because zero through 3 does have
the capability of human interaction. So I'll borrow a phrase
from my colleague here that when we talk about artificial
intelligence in that 4 and 5, it really does go far beyond what
humans can do.
So when I look at AVs myself, and I take out my role as
National President, and I look at my role as a parent or as a
mom, and I look at AVs--and, again, just relating myself to the
4s and 5s--there needs to be a cure to stop 35,000 deaths
that--not only drunk driving deaths, but there are also
behavioral traffic-related deaths--that we have the opportunity
to have a cure here, and the AVs themselves, the 4s and 5s,
could actually be that vaccine that we could utilize across the
country to save more lives. So I'd have to answer it, in terms
of self-driving, I'm really looking at 4 and 5 as being that
vaccine over time.
Senator Blumenthal. And do you think it's important that
NHTSA make sure that certain safety features are mandatory in
all vehicles so that drivers and individuals know what to
expect, not only for their sake, but for the sake of people who
are driving on the road unaware about the lack of those safety
features?
Ms. Sheehey-Church. Oh, absolutely. I mean, just like with
seatbelts, when we needed to educate, it's like we're educating
on ignition interlocks right now. So now the consumer
acceptance will only come with pure education on what AVs can
do. So it's really--I beg that NHTSA and the auto industry and
safety advocates really sit at the table and all have a
collaborated effort in bringing the AVs--we all want the same
thing. We want safety. Safety comes first. So when I say that,
you know, before anything is rolled out in masses, that safety
is our number one concern, and every one of these individuals
in the auto industry, safety advocates, have to be at that
table to be able to collaborate together.
Senator Blumenthal. Mr. Maddox, how do we make sure that
voluntary standards regarding safety are adopted?
Mr. Maddox. I think the key there is that we enter into the
activity to create those voluntary standards with a number of
outcomes in mind, certainly one of them being the creation--the
support for the creation of FMVSS. So if you look at the
history of safety technology over the decades, every major
safety technology on our vehicle today started with that
voluntary standard approach.
So if we use that, there's no reason we can't continue that
process that works for automated vehicles. It is much more
complex. As Mr. Csongor has pointed out, the data involved here
is far, far higher--in order of magnitude, higher than prior
safety technologies. So the key is for NHTSA, U.S. DOT, the
auto industry, and other stakeholders to really get together
and decide what's the end game, what voluntary standards would
be needed to enable the technology to be deployed at scale,
and, very specifically, which pieces of those voluntary
standards would inform future Federal regulations.
Senator Blumenthal. Thank you. Thank you all for being
here. I appreciate your excellent testimony today.
The Chairman. Thank you, Senator Blumenthal.
Senator Klobuchar?
STATEMENT OF HON. AMY KLOBUCHAR,
U.S. SENATOR FROM MINNESOTA
Senator Klobuchar. Thank you very much. Thank you for being
here.
I wanted to thank you, specifically, for your work, Ms.
Sheehey-Church. As you know, as a former prosecutor, I did a
lot of work with MADD, and we passed finally in Minnesota the
felony DWI law, which was helpful, and I am supportive of your
words about how technology can help as well.
Mr. Bainwol, I think I would start with you. As you know,
fully autonomous vehicles are still a few years out. What
technology is on the road now to intervene if a driver is
distracted behind the wheel?
Mr. Bainwol. Good question. There are----
Senator Klobuchar. It's hard to find questions to ask after
everyone has asked them. So I really appreciate that. Thank
you.
Mr. Bainwol. Yes. So you did a good job. So, generally,
crash avoidance technologies are the answer to your question.
Automatic braking is probably the best example, where the car
will brake for the driver if the driver is not engaged soon
enough. Drivers should not drive in a distracted situation. We
shouldn't rely entirely on the car to make up for human error.
There's a human responsibility here, and there's no
question that, as a society, we can do the same kinds of things
we did with MADD, the success of MADD, that kind of changed the
culture. We need to do the same with distraction. But there are
technologies that will help, and emergency braking is probably
the best example.
Senator Klobuchar. I think so.
Mr. Maddox, as we see these changes coming upon us, what
are some of the steps that can be taken to ensure consumers
understand this with some of the current changes that are
there? I just picture people running cars and not used to it or
people getting a new car. I know I had the same car for--now my
car confessions occur--for 12 years, a Saturn. It was great,
and then it got a hole where I put the brake down in the
actual----
[Laughter.]
Senator Klobuchar. So I got a new car, which I truly enjoy,
a Malibu, and it has some of the brakes--if you go over the
lane a little bit, it sort of jerks you back, and at first,
even though they told me about it, it was a surprise. So I'm
just picturing some of this. How do you educate consumers?
Mr. Maddox. I think that is another great question.
Certainly, I would commend--NHTSA has undertaken a program. I
think it's called My Car Does What?, and it's actually being
implemented----
Senator Klobuchar. That seems like a perfect thing for me.
Mr. Maddox. Exactly, I think, for a lot of us, as a matter
of fact, even us experts. We don't get to drive every car with
every little nuance of technology. So I think that educating
the consumers on what their vehicle does and what it doesn't
do----
Senator Klobuchar. And the limits as well. Senator Hoeven
and I have led a lot of efforts on distracted driving and got a
bill passed just recently with Senator Thune's help. Thank you.
It was part of the FAST Act, which allowed some of the grants
to go out for education efforts to the states on distracted
driving. For a while, only one state was getting the money:
Connecticut.
[Laughter.]
Senator Klobuchar. So I think this will have to be a major
part of the effort.
Mr. Maddox. Yes. In fact, I would add that with the
automated vehicle proving ground program, we are looking at how
to utilize those 10 sites to do that consumer education.
Imagine--what we're considering in Michigan is that on the
second Sunday or the fourth Sunday in every month, we would
invite the public in and let them actually experience this
technology in a vehicle--maybe they don't own it yet--because,
really, driving it and experiencing it is worth a thousand
words and pictures.
Senator Klobuchar. Let me just go to a few quick questions
for you, Mr. Maddox.
Mr. Maddox. Sure.
Senator Klobuchar. Ensuring that highway construction
workers are safe in a work zone is critical to the testing and
deployment of connected automated vehicles. Since work zones
often temporarily block off lanes and redirect traffic flow,
they can be difficult to navigate, even for experienced
drivers. Do you believe it would improve safety if temporary
traffic control devices could send safety related information
to the sensors on the vehicles?
Mr. Maddox. Absolutely. In fact, it's a great V to I
application. The V to I and V to V really are critical for
automated vehicles, also.
Senator Klobuchar. OK. Thank you. Last question. Senator
Hoeven and I also worked on a bill to make it clear that the
owner of a vehicle also owns any information collected by an
onboard event data recorder. It was signed into law also as
part of the FAST Act.
Mr. Maddox, connected vehicles will collect and share large
amounts of data. What protections should be in place to keep
consumers' personal information secure?
Mr. Maddox. I can't say that I've got a comprehensive
answer to that question at this point. I do believe that we do
need to establish rules on that data ownership, and it does
seem that the owner or the operator of the vehicle should be
the owner of the data.
Senator Klobuchar. And, obviously, for emergencies and for
crash sites and things like that, you would want more
information to go to the authorities.
Thank you very much, and I appreciate all of you.
Senator Peters [presiding]. Senator Cantwell?
STATEMENT OF HON. MARIA CANTWELL,
U.S. SENATOR FROM WASHINGTON
Senator Cantwell. Thank you, Mr. Chairman, and thanks to
the witnesses.
This industry, I think, is estimated to generate something
like $7 trillion by 2050, so we certainly want to continue to
see the innovation and are proud that Washington and Kirkland,
Washington, specifically, have also been some of the early
adopters to enabling this kind of testing.
I wanted to ask you, Mr. Csongor, about working with the
trucking industry, particularly PACCAR. You know, from the
aviation side, these efficiencies have got pilots doing very
little manual operation of our airplanes, and these kinds of
innovations can also reduce the demand on drivers. But the
obvious efficiencies that we need to get out of freight is very
important for my state, very important for the movement of
freight.
What do you think that this will do to help us on
efficiencies?
Mr. Csongor. That's a great question. We actually, I think,
as you may know, are working with PACCAR, and we had actually
shown a video and demonstrated a PACCAR truck running on
basically a computer very similar to this one. So the truck is
able to drive and maintain in the lanes, and the deployment of
this product is up to PACCAR and to decide how they want to do
it and what the policy is. But the capability, certainly, to
reduce driver fatigue and to assist the trucker is certainly
there and could be deployed fairly soon.
Now, it's very hard to predict the evolution of jobs with
the introduction of this technology. The role of a pilot has
changed, but the job remains. So with regard to the trucking
industry, I think the most immediate benefit would be trucker
fatigue. But then the actual implementation of it and how it
affects it I would have to defer to someone from that industry.
Senator Cantwell. What about fuel efficiency? That's not in
the target?
Mr. Csongor. Oh, it is. As you can imagine, having a
computer driving, it would obviously be able to drive more
efficiently than a human would and do it with very little
effort. So I think that there is also financial benefits to the
truckers beyond safety.
Senator Cantwell. Fuel efficiency for us in the Northwest
is a very big issue because we have some of the highest gas
prices in the Nation as our supply comes from Alaska and is a
somewhat limited market. Anything that helps us drive
efficiencies on fuel is very, very important for us in
maintaining what is one in three jobs related to trade.
Being a big importer of a lot of products, moving them
through the country cost effectively or moving agricultural
products out of the Midwest through our ports--all of this is
about reducing fuel costs and efficiencies to make operations
more successful. If we don't, it goes to Canada or somewhere
else. I don't know if anybody else, perhaps Mr. Maddox, wanted
to comment on that.
Mr. Maddox. I would like to comment on that, actually. I
totally agree with you. You know, the cost of moving goods,
especially in a large geographic area of the United States,
really is critical. It's a critical economic issue,
competitiveness issue for us. Automated and connected
technology can enable truck platooning, which brings
significant fuel economy and potential savings, and up even in
the 10 percent range. That's a number unheard of with most
other vehicle efficiency technologies that are on the near
horizon.
Senator Cantwell. You want to elaborate on platooning a
little bit?
Mr. Maddox. I'm sorry. My apologies. Platooning is where
one vehicle follows another very closely. In the case of a
truck, it could be anywhere as short as 15 feet but probably
more like 25 to 40 feet behind the leading truck. Automation
with connection can enable that technology with or without a
driver in the lead--certainly in the lead vehicle or the
following vehicles.
The key point, though, is the efficiency benefit, if it's
the following trucks, as well as the leading truck. So any
technology that could deliver a 10 percent fuel savings has to
be examined in great detail, especially for a country the size
of the United States.
Senator Cantwell. Well, thank you for bringing that up. I
really do think it is about competitiveness, and I know we have
many challenges. I know here, we're always saying, ``This is
Washington and we want to regulate,'' and I keep saying, ``I'm
from Washington and we want to innovate.'' And the innovation
aspect of this is critically important, so I hope our state
continues to innovate in this area.
Thank you.
Senator Peters. Thank you, Senator Cantwell.
Senator Udall?
STATEMENT OF HON. TOM UDALL,
U.S. SENATOR FROM NEW MEXICO
Senator Udall. Thank you, Senator Peters, very much, and I
thank the panelists today. I think this has been a very
productive discussion we've had here.
I'm particularly interested in autonomous and semi-
autonomous vehicles because Rio Rancho, New Mexico, is home to
a major Intel semiconductor fabrication plant, and Intel's
facility is an economic anchor for that city and central New
Mexico as a whole, providing high-paying jobs to New Mexico
STEM graduates. This facility in Rio Rancho, this Intel
facility, makes some of the semiconductors used in modern high-
tech vehicle technology. So I'm going to be following this very
closely and look forward to, if there are additional questions,
submitting some to the record.
But my first question is to Ms. Sheehey-Church and has to
do with the safety aspects of autonomous vehicles. This
technology has the promise to reduce crashes and fatalities, as
you've talked about a little bit here, including those from
drunk driving. As you know, I've worked with MADD for many
years, going back to my service as Attorney General in New
Mexico. We had a comprehensive package that we put in, working
with MADD, that really reduced drunk driving and drunk driving
deaths in New Mexico.
I'm sure that MADD is looking past the industry hype and
enthusiasm and is focused on safety and saving lives. So while
the benefits of self-driving technology for drunk driving are
obvious at first blush, we must remember that the intoxicated
person could be sitting at the wheel of an autonomous vehicle
and could take over at any time, especially if there's an issue
that requires human intervention. Some states have proposed
laws on this topic.
Should an autonomous vehicle be required to have technology
to prevent an intoxicated person from taking control of the
wheel?
Ms. Sheehey-Church. Senator, that's a great question, and I
don't know if I have the full answer to it, whether they will
or will not have that. I know that DADSS in itself in AV
vehicles, the 4 and 5 levels, are great complements to each
other, but in terms of digging deeper into the AV
technicalities, I'm not a technical expert. I only play one on
TV. So I can't tell you what exactly is going to be in that.
I'd have to probably defer to one of my colleagues to see if
that was going to be. But I'm in full agreement that if that
individual can take over at any time, that is a concern,
obviously, for MADD.
Senator Udall. Yes. And you mentioned DADSS, and that's a
very important part of this as developing technology----
Ms. Sheehey-Church. Absolutely.
Senator Udall.--so that vehicles won't be turned over to an
intoxicated person. Now, what are your key recommendations for
Congress and Federal agencies to ensure that this technology
develops so we can realize the safety potential of autonomous
vehicles to eliminate drunk driving deaths?
Ms. Sheehey-Church. I go back, Senator, to the four
principles that I said in my written testimony. It is really
looking at and supporting the Federal regulatory framework.
There's the need for that, and also supporting the existing
state systems that we have. See, you're not listening and
that's why you're tapping him on the shoulder, because I would
do that if I was up there.
[Laughter.]
Senator Udall. No----
Senator Markey. We're both listening.
Senator Udall. Yes. We're both listening.
Ms. Sheehey-Church. Just kidding. Just kidding. And,
really, the support on MADD's side for Levels 4 and 5 is really
looking at that in terms of the technologies, but also really
supporting and evaluating the technology and the data that we
need to get, evaluating the data as this whole thing evolves. I
know we've got skeptics on the fence. We are not going to
dismiss that. But I think the more that NHTSA, the governing
bodies, the auto industry, and the safety advocates can sit at
the same table, we can start to overcome those, and, obviously,
from MADD's point of view, we'd like to overcome them sooner
rather than later.
Senator Udall. Do any of the other panelists have comments
on the safety issue?
Mr. Csongor. The safety issue, I think, is very important,
as we've talked about, and I very much agree with what she
said. The safety issue on both protection of the data as well
as protection of the system is something that I think will be
solved, like we've solved other computer problems--a
combination of industry standards and the right regulations
that help support them, and I think working together, and I
think the devil is in the details.
Senator Udall. Thank you very much.
Senator Peters. Thank you, Senator Udall.
Senator Markey?
STATEMENT OF HON. EDWARD MARKEY,
U.S. SENATOR FROM MASSACHUSETTS
Senator Markey. Thank you, Mr. Chairman, very much.
Ms. Sheehey-Church, we were--believe me, I was listening
very intently.
Ms. Sheehey-Church. I'm only kidding. It's that kind of--
like a squirrel, I'm looking at anybody.
Senator Markey. No, no, I know it. I was run over by a car
when I was five years old, and nothing will focus you so much
on an issue like vehicle safety as event like that. So I thank
you for all your great work, historic work, on these issues
over the years.
Autonomous vehicles are just computers without drivers.
That's all they are. And as we move to the future, we then have
to ensure that we are going to build in all of the human values
that we want to see built into these new computers that will be
riding around on the streets of America.
Senator Blumenthal and I, in 2013 and again last year,
asked 20 automakers what they're doing to protect these
computers on wheels. Here's what we learned. We learned that
thieves no longer need a crowbar to break into your car. They
just need an iPhone. A few years ago, we witnessed firsthand
how easily cars can be hacked. We watched as hackers remotely
took control of the brakes, the steering, and the acceleration
of a Jeep Cherokee. Chrysler had to recall 1.4 million vehicles
to fix this cybersecurity problem.
Rather than addressing the cybersecurity problems after a
hack has occurred, we must ensure that robust cybersecurity
protections are built into the design, the construction, and
operation of these transportation technologies, particularly as
we move into this new era very quickly. That's why Senator
Blumenthal and I have introduced legislation once again, the
Spy Car Act, that directs NHTSA and the Federal Trade
Commission to establish Federal standards to secure our cars
and to protect our privacy.
We should not have to choose, as Americans, between being
connected and being protected. This should not be a tension.
These issues should all be solved before the new era unfolds.
So, if you could, each panelist, just give me your answer as to
whether or not you believe cars should have mandatory
cybersecurity standards, including hacking protections that
protect all access points in a car; data security measures that
prevent unwanted access to all collected information; and
hacking mitigation technologies that can detect, report, and
stop hacking attempts in real time. Should that be a
requirement so that these hackers can't get into the system?
Mr. Bainwol, yes or no? Should we have those standards?
Mr. Bainwol. The short answer is no. The longer answer is
we share your concern and we share your objective, but we think
that the best way to realize your objective is to have a
dynamic approach, and our fear is that standards would become
obsolete very quickly.
Senator Markey. I appreciate that, but I don't agree with
you.
Mr. Csongor--you need dynamic mandatory standards. You're
right. The standard could keep getting raised, but to have no
standard would be extremely dangerous. That's the world I grew
up in, with no seatbelts, no airbags, the steering wheel made
out of metal, and so that just can't be the standard as we move
into this computer world, from my perspective.
Mr. Csongor?
Mr. Csongor. Senator Markey, I'd certainly agree with the
goal that you're describing. I'm not sure about the
implementation of it, particularly when you talk about the
ability of the car to be able to heal itself or to be able to
solve the problem. We don't know what we don't know, and
cybersecurity is not something that we fix once and then it's
done.
Senator Markey. Exactly.
Mr. Csongor. It's something that you're battling
continuously.
Senator Markey. Exactly. And should the industry be
mandated to be fighting it continuously? Should that be part of
the NHTSA rules?
Mr. Csongor. Right. I think the only question is just
what's the right balance and----
Senator Markey. Right. But the balance should be that it
can't be hacked and that the constant high standard that the
industry has to meet is one that protects against that. Do you
think that's a reasonable goal to set?
Mr. Csongor. I think that's a reasonable goal.
Senator Markey. Thank you.
Let me ask you, Ms. Sheehey-Church. Do you agree with that?
Ms. Sheehey-Church. Senator, I think it's a reasonable
expectation, but in the position that I am and with the staff
that we work with as we work with the panelists, I'm not at
this time able to answer the question on what could be put on
to the NHTSA for regulations on hacking. I appreciate the
question.
Senator Markey. I got you.
Ms. Sheehey-Church. I think it's a serious concern, but I
can't answer it at this time.
Senator Markey. Mr. Maddox?
Mr. Maddox. I think the nature of the risk for
cybersecurity is changing so quickly, voluntary standards,
frankly, have a better opportunity to keep up with that
changing risk than any regulation could. Second, I do believe
that we need to think about transportation cybersecurity as a
system--security in our vehicles, security in our traffic
control, security in our signs, security even in the
communication systems that we rely on increasingly for
transportation. So I think voluntary standards would likely
have a more effective result.
Senator Markey. I think the history, though, shows that
with airbags and with seatbelts, et cetera, unless there's a
mandate, it's just not accomplished unless you actually pass
laws out of this committee nationally. The industry moves
slowly, and so the best players move voluntarily. The worst
players don't, and the worst players are the ones that cause
all the damage out on the highways. So you need to have, you
know, those protections so that there is that minimum standard.
And I think that, you know, just as part of this
conversation, it's good as we start in this era that we begin
to have this discussion, because to a certain extent, there's a
repetition syndrome, where the industry is going to be
promoting all the wonderful aspects of it, but there's a
Dickensian aspect to the Internet, which is what we're talking
about here, just to talk about computers. It's the best of
technologies and it's the worst of technologies. It's a
technology that can enable and ennoble and it's a technology
that can degrade and debase and harm families. So it's both at
the same time.
Talking about all the great things has to be accompanied by
what we're going to do to minimize the bad things that can
happen. By ignoring that discussion, we ignore the central
concerns that families will have about safety and there will be
a lot of concerns families are going to have about autonomous
vehicles and the safety of their children in those vehicles.
From my perspective, what I want to hear in the weeks,
months, and years to come is that any standards would require
computers on wheels to constantly update and patch themselves
for any vulnerabilities. I believe that has to be mandatory.
You can't just leave it up to any one auto manufacturer to do
it. You have to have every one of the players accepting that as
a responsibility. Otherwise, the streets won't be safer. These
vehicles will be very dangerous. These vehicles will be hacked.
These vehicles will subject families to the kind of tragedies
that we're trying to protect against.
So you have to build in both simultaneously. I mean, it's
the best and the worst, you know, in any technology, and each
technology is only as good as the values which we, as a people,
animate those technologies with. How high of a standard do we
want to make sure that we protect families against the downside
of that technology. And, by the way, that's every technology.
None of them are exempt from that kind of a situation.
That will be my concern going forward, because I obviously
want to see a revolution occur, but not with kind of
panglossian glasses that ignore the fact that there are
vulnerabilities. I know what happens when those safeguards are
not in place.
Thank you, Mr. Chairman.
Senator Peters. Thank you, Senator Markey.
Before we close out the hearing, I need to take care of a
little housekeeping, if you could bear with me. I would ask
unanimous consent to include in the record letters from
stakeholders providing additional perspective on today's
hearing, including a statement from the Advocates on Highway
and Auto Safety, the American Car Rental Association,
Continental, the Property Casualty Insurers Association of
America, and Uber Technologies, Incorporated. Hearing no
objection.
[The information referred to can be found in the Appendix.]
Senator Peters. One final note. The hearing record will
remain open for 2 weeks. During this time, Senators are asked
to submit any questions for the record. Upon receipt, the
witnesses are requested to submit their written answers to the
Committee as soon as possible.
So in closing, I just want to thank our four witnesses for
what I think was incredible testimony dealing with incredible
technology that will surely be transformative. Thank you for
your time today. We look forward to working with you in the
years ahead.
With that, this hearing is now adjourned.
[Whereupon, at 12 p.m., the hearing was adjourned.]
A P P E N D I X
Prepared Statement of Jacqueline S. Gillan, President,
Advocates for Highway and Auto Safety
Introduction
Advocates for Highway and Auto Safety (Advocates) is a coalition of
public health, safety, and consumer organizations, insurers and
insurance agents that promotes highway and auto safety through the
adoption of Federal and state laws, policies and regulations. Advocates
is unique both in its board composition and its mission of advancing
safer vehicles, safer drivers and safer roads. We respectfully request
that this statement and the comments Advocates submitted to the public
docket in response to the National Highway Traffic Safety
Administration (NHTSA) ``Federal Automated Vehicles Policy'' (AV
Guidelines) Notice and Request for Comments,i which are
attached, be included in the hearing record.
---------------------------------------------------------------------------
\i\ 281 F.R. 65703 (Sept. 23, 2016); DOT Docket No. NHTSA-2016-
0090.
---------------------------------------------------------------------------
Motor Vehicle Deaths are Climbing
According to the Federal Government, each year motor vehicle
crashes kill tens of thousands and injure millions more at a cost to
society of over $800 billion.ii Unfortunately, deaths
resulting from motor vehicle crashes are on the rise. According to
NHTSA, in 2015 our Nation experienced the largest percentage increase
of motor vehicle deaths in nearly fifty years.iii More than
35,000 people were killed on our Nation's roads, representing a 7.2
percent upturn.iv Preliminary information for the first nine
months of 2016 appears to be even worse, indicating an 8 percent rise
in fatalities compared to the same time period in 2015.v
Advocates firmly believes that automated vehicle (AV) technology has
the potential to make significant and lasting reductions in this
mortality and morbidity toll.
---------------------------------------------------------------------------
\ii\ The Economic and Societal Impact of Motor Vehicle Crashes,
2010 (Revised), HS 812 013, U.S. DOT, NHTSA (May 2015 (Revised)),
available at http://www-nrd.nhtsa.dot.gov/Pubs/812013.pdf. (NHTSA Cost
of Motor Vehicle Crashes Report).
\iii\ National Center for Statistics and Analysis, 2015 motor
vehicle crashes: Overview, Report No. DOT HS 812 318, National Highway
Traffic Safety Administration (Aug. 2016).
\iv\ Id.
\v\ National Center for Statistics and Analysis, Early Estimate of
Motor Vehicle Traffic Fatalities for the First 9 Months of 2016, Report
No. DOT HS 812 358, National Highway Traffic Safety Administration
(Jan. 2017).
---------------------------------------------------------------------------
Advocates Has Consistently Promoted Advanced Technologies in Vehicles
to Save Lives and Prevent Injuries
Advocates has always enthusiastically championed vehicle safety
technology and for good reason. It is one of the most effective
strategies for preventing deaths and injuries. NHTSA has estimated that
since 1960, over 600,000 lives have been saved by motor vehicle safety
technologies.vi In 1991, Advocates led the coalition that
supported bipartisan legislation sponsored by former Senators John
Danforth (R-MO) and Richard Bryan (D-NV) that included the airbag
mandate in the Intermodal Surface Transportation Efficiency Act (ISTEA)
of 1991.vii As a result, by 1997, every new car sold in the
United States was equipped with a front seat airbag and the lives saved
have been significant. In fact, airbags save over 2,000 lives
annually.viii
---------------------------------------------------------------------------
\vi\ Lives Saved by Vehicle Safety Technologies and Associated
Federal Motor Vehicle Safety Standards, 1960 to 2012, DOT HS 812 069
(NHTSA, 2015); See also, NHTSA AV Policy, Executive Summary, p. 5
endnote 1.
\vii\ Pub. L. 102-240 (Dec. 18, 1991).
\viii\ National Center for Statistics and Analysis, Lives Saved in
2015 by Restraint Use and Minimum-Drinking-Age Laws, National Highway
Traffic Safety Administration, Report No. DOT HS 812 319 (Aug. 2016).
---------------------------------------------------------------------------
Advocates continued to build on this success by supporting
additional lifesaving technologies as standard equipment in all
vehicles in other legislation and regulatory proposals. These efforts
included: tire pressure monitoring systems;ix rear outboard
3-point seat belts;x electronic stability
control;xi rear seat belt reminder systems;xii
rear view cameras;xiii brake transmission
interlocks;xiv seat belts on motorcoaches;xv
electronic logging devices;xvi and, crash avoidance systems
such as automatic emergency braking. These safety advances have saved
hundreds of thousands of lives and have been accomplished because of
the bipartisan leadership of the Members of the Senate Commerce,
Science and Transportation Committee.
---------------------------------------------------------------------------
\ix\ Transportation Recall Enhancement, Accountability, and
Documentation (TREAD) Act, Pub. L. 106-414 (Nov. 1, 2000).
\x\ Anton's Law, Pub. L. 107-318 (Dec. 4, 2002).
\xi\ Safe, Accountable, Flexible, Efficient Transportation Equity
Act: A Legacy for Users (SAFETEA-LU), Pub. L. 109-59 (Aug. 10, 2005).
\xii\ Id.
\xiii\ Cameron Gulbransen Kids Transportation Safety Act of 2007,
Pub. L. 110-189 (Feb. 28, 2008).
\xiv\ Id.
\xv\ Moving Ahead for Progress in the 21st Century (MAP-21) Act,
Pub. L. 112-141 (Jan. 3, 2012).
\xvi\ Id.
---------------------------------------------------------------------------
NHTSA Has a Statutory Duty to the Public to Ensure the Safety of
Autonomous Vehicles
Fifty years ago, Congress passed the National Traffic and Motor
Vehicle Safety Act of 1966 because of concerns about the death and
injury toll on our highways.xvii The law required the
Federal Government to establish Federal motor vehicle safety standards
(FMVSS) to protect the public against ``unreasonable risk of accidents
occurring as a result of the design, construction or performance of
motor vehicles.'' xviii While motor vehicles have changed
dramatically since that time and will continue to do so in the future,
the underlying premise of this prescient law and NHTSA's safety mission
have not.
---------------------------------------------------------------------------
\xvii\ Pub. L. 89-563 (Sept. 9, 1966).
\xviii\ Title 49, U.S.C. Sec. 30102.
---------------------------------------------------------------------------
Unfortunately, NHTSA has chosen to issue only ``voluntary
guidelines'' for the development of AVs. Voluntary guidelines are not
legally binding, are unenforceable and, therefore, are inadequate to
ensure safety and protect the public. Manufacturers may choose to
deviate from the guidelines or ignore them entirely at any time and for
any reason including internal corporate priorities such as cost or
marketing considerations. In addition, some entities may choose to
follow the guidelines while others may not, creating a dangerous
patchwork of noncompliance. Consumers and NHTSA also have no legal
recourse against a manufacturer's failure to follow the guidelines.
NHTSA cannot bring an enforcement action, force a statutory recall, or
even influence a voluntary recall for failure to abide by the
guidelines.
A Functional Safety Approach is Essential to Provide the Framework for
the Design, Development and Deployment of Autonomous Vehicle
Technology
Before the widespread introduction of AV technology to the
commercial marketplace and deployment on public roads, NHTSA must
establish basic safeguards to protect the public. Advocates strongly
recommends that AV manufacturers be required to meet a ``functional
safety standard'' to guarantee safety to the maximum extent feasible of
the overall system performance. While manufacturers will still have to
certify vehicles meet all applicable FMVSS that address the mechanical
operation of vehicle safety systems, a functional safety standard would
apply to the AV operating system to ensure that the controlling
software performs as designed.
Functional safety is a well-known process by which a product is
designed, developed, manufactured and deployed to ensure that the
product, as a whole, will function safely, as intended, and will
prevent or mitigate misuse which could lead to unsafe conditions. A
similar process is currently utilized by the Federal Aviation
Administration (FAA) in portions of its regulations to ensure safety
while encouraging innovation.xix
---------------------------------------------------------------------------
\xix\ See, e.g., 14 USC Subpart F--Equipment, Sec. Sec. 25.1301,
Function and installation, and 25.1309, Equipment, systems, and
installations.
---------------------------------------------------------------------------
Additionally, the submission of a Safety Assessment Letter (SAL)
must be mandatory and not at the whim or choice of a manufacturer.
There is too much at stake to allow optional and discretionary
adherence. Under a functional safety standard, a manufacturer must
certify to NHTSA through the mandatory submission of a SAL that the AV
has been tested to ensure that it will operate properly and safely
under the conditions the vehicle is designed to encounter (i.e.,
congestion, weather and road conditions, human/machine interface and
vehicle interaction with other road users). The mandatory submission of
the SAL should include all information required by the Secretary of
Transportation including test results and data to support the
conclusion that the AV system functions as designed. This submission
should also include any negative test results and data that may
indicate the AV system did not always function as designed along with
information and data indicating how these issues were resolved by the
manufacturer. In addition, any and all documentation regarding how
cybersecurity threats were addressed should be included.
Prior to introduction of the AV system into the stream of commerce,
NHTSA would review the SAL, test results and data, and then consult
with the manufacturer. The manufacturer would be required to respond to
any questions or concerns from the agency. This process is intended to
allow NHTSA to bring to the manufacturer's attention any issues and/or
problems the agency identifies in the SAL information, tests results
and data supplied by the manufacturer prior to introduction of the AV
system into the marketplace. This process would provide NHTSA with
technical information about products coming to market and provide AV
manufacturers with technical feedback in an efficient and effective
process. The SAL and the questions raised by NHTSA, however, are not
intended as pre-market approval. The manufacturer, after having
responded to the agency's inquiries, may introduce the AV system into
the stream of commerce.
Additional Authorities and Resources are Critical
Regulating AVs presents unique challenges for NHTSA, and those
issues warrant the agency being given additional tools to protect
against potentially catastrophic defects. Flaws or viruses in computer
software of AVs could adversely affect thousands of vehicles
simultaneously. The NHTSA, therefore, must be given imminent hazard
authority in order to expedite the grounding of vehicles that the
agency has identified as having a potentially dangerous software
problem that could lead to crashes, deaths and injuries. Also, because
of the potential serious nature of any software problem that could
affect thousands of vehicles and result in deaths and injuries, the
ability to levy criminal penalties is essential. Criminal penalties
will deter manufacturers and suppliers from knowingly and willfully
permitting the manufacture and sale of AV systems with flawed software
operating systems that could pose a danger to human life in the event
of a crash.
Providing further broad statutory exemptions from the FMVSS for AVs
is both unnecessary and unwise. There is already a statutory process in
place for manufacturers to seek an exemption from the FMVSS that
Congress enacted only two years ago. Pursuant to Section 24404 of the
Fixing America's Surface Transportation (FAST) Act xx
codified at 49 USC Section 30112(b)(10), manufacturers of AVs are
permitted an unlimited number of vehicles that can be exempt from one
or more of the FMVSS for testing or evaluation. Exempt vehicles under
this provision may not be sold or resold to the public. Furthermore,
under 49 USC Section 30113 a manufacturer may receive an exemption from
compliance with the FMVSS for the sale of a vehicle for not more than
2,500 vehicles to be sold in the United States in any 12-month period.
Until a functional safety standard is applied to AVs, this cap should
not be raised from its current level.
---------------------------------------------------------------------------
\xx\ Pub. L. 112-141 (Dec. 4, 2015).
---------------------------------------------------------------------------
Manufacturers may seek appropriate exemptions under the current
process until NHTSA revises the FMVSS for level 4 and 5 AVs. However,
while level 4 and 5 AVs may be exempt from parts of certain FMVSS or
other regulations, much of the performance standards for safety systems
would still apply. For example, even if a brake pedal is not needed for
AV control, the rest of the brake system requirements in FMVSS 135 will
still be needed to ensure the AV controls can stop the vehicle within
the required stopping distance. Even for level 4 and 5 AVs,
manufacturers will still have to certify to most performance
requirements of the existing FMVSS. However, because level 2 and 3 AVs
will still require significant operation by a human driver, these
vehicles should not be exempt from compliance with the current body of
FMVSS and regulations.
Vehicle and technology companies are already putting some AVs out
on public roads and beginning to market these systems to the public.
The development of AV technology is not just taking place in the United
States. In fact, AVs are being tested throughout the globe in places
such as the United Arab Emirates and Singapore.xxi Vehicles
imported from overseas have to meet the requirements of the FMVSS.
However, if only voluntary guidelines are in place for AVs, foreign
companies and entities may export products that are dangerously unsafe
and NHTSA will have little recourse. In addition, some in the motor
carrier industry have predicted that automated technology will be
placed in trucks before passenger vehicles.xxii The
potential for an 80,000 pound rig using untested and unregulated
technology on public roads is a very real scenario if NHTSA continues
to merely rely on voluntary guidelines and the ``good intentions'' of
manufacturers introducing automated technology.
---------------------------------------------------------------------------
\xxi\ 2025 AD Newsletter, MIDDLE EAST COMBATS ROAD DEATHS WITH
DRIVERLESS CARS (Nov. 2016) available at: https://www.2025ad.com/the-
week-in-ad/arab-emirates-driverless-cars/; Andrew Hawkins, One of
Europe's largest carmakers will test its self-driving cars in
Singapore, The Verge (May 3, 2017).
\xxii\ Seth Clevenger and Eric Miller, ATA's Chris Spear Calls for
Industry to Embrace Technology, Transport Topics (Mar. 2, 2017).
---------------------------------------------------------------------------
It is essential that NHTSA immediately adopt a functional safety
standard to minimize public exposure to unreasonable risks of motor
vehicle crashes involving these experimental systems. For example, the
fatal crash of a Tesla Model S in Florida in May 2016, could have been
averted had a functional safety standard required due diligence testing
of the Autopilot System prior to deployment. The driver was using the
Tesla Autopilot system when the vehicle passed under the side of a
truck trailer that was turning across the highway, resulting in the
fatality of the driver. After the crash, Tesla admitted to NHTSA that
it had considered misuse of the Autopilot system including distracted
driving and the use of the system outside preferred environments and
conditions. Under a functional safety standard, if the misuse was
foreseeable, Tesla would be required to address the misuse prior to
confirmation to NHTSA that the vehicle was safe.
Recommendations:
NHTSA must require that manufacturers meet a ``functional
safety standard'' to guarantee safety of AVs before they are
introduced into the marketplace.
Manufacturers must be required to submit a Safety Assessment
Letter (SAL) that confirms that the AV has been tested to
ensure it operates safely. The SAL should include all of the
test results and data to support the conclusion that the AV
system functions as designed.
NHTSA should be given the additional tools of imminent
hazard authority to protect against potentially catastrophic
defects with AVs and criminal penalties to ensure manufacturers
do not willfully mislead or misinform the agency.
Providing further broad statutory exemptions from the FMVSS
for AVs is both unnecessary and unwise. Until NHTSA revises the
FMVSS for level 4 and 5 AVs, manufacturers may seek appropriate
exemptions under the current process.
Autonomous Vehicles Must Have Adequate Cybersecurity and Privacy
Standards to Protect the Public and Must Share Information on
Critical Events
A failure to adequately secure AV systems and to protect against
cyber-attacks could endanger AV passengers, non-AV motorists,
pedestrians, bicyclists and other vulnerable roadway users. It could
also clog roads, stopping the movement of goods and hindering the
responses of emergency vehicles. NHTSA should identify cybersecurity
problem areas and require specific responses from manufacturers as to
how those are being addressed. Problem areas could include subjects
such as global position system (GPS) signal loss or degradation,
spoofing, and off-line and real time hacking of single vehicles or
fleets of vehicles. AV cybersecurity should be tested as part of the
functional safety standard to ensure that the vehicle cybersecurity
system operates as designed. The potential and real risk of a
malevolent computer hack impacting hundreds or thousands of AVs,
perhaps whole model runs, makes strong cybersecurity protections a
crucial and essential element of AV design.
Additionally, data sharing among manufacturers is essential to
improve overall safety among AVs. Data and information about known
flaws or problems encountered during development and while in use must
be shared among manufacturers and with NHTSA and the public to ensure
that all AV systems are learning about problems in real time and can
benefit from the experience of other AV systems. This type of
collaborative development is already taking place in the industry with
the creation of the Automotive Information Sharing and Analysis Center
(ISAC). Data sharing will expedite solutions to unusual or unique
safety problems and ensure they are readily identified and corrected.
Similarly, AVs should be subject to binding privacy standards to
ensure that data is not abused. The recording and sharing of data will
be critical to achieving the safest performance of AVs. Keeping the
public informed as to the importance of sharing data and ensuring their
privacy will also be critical to ensure public participation and
acceptance. The industry and regulators must guarantee that data is
protected and only used for the purposes of improving safety, and not
for other commercial uses, such as the marketing of products based on
vehicle location, which could turn the public against data sharing.
Recommendations:
AVs must have adequate cybersecurity and privacy standards
to protect the public.
AVs should be subject to binding privacy standards to ensure
that consumer data is not used for purposes other than
improving the safety and security of AVs.
The Development of Autonomous Vehicles Must Be Transparent or Public
Confidence in the Technology Will Suffer
The development and deployment of AVs as well as NHTSA's role in
regulating this technology must be open and transparent. All non-
propriety communications and responses between the agency and a
manufacturer as it relates to any issues involving AVs must be made
available for public review and scholarly research. All data generated
from the testing and deployment of AVs, except for trade secrets and
private individual information must also be made public. Lack of
transparency will severely undermine the public's confidence in this
new technology and inhibit its widespread adoption.
Opinion polls already show strong public skepticism and hesitation
about AVs and those doubts are surely warranted. Over the last few
years, automakers have hidden from the American public and regulators
safety defects that have led to numerous unacceptable and unnecessary
deaths and injuries as well as the recall of tens of millions of
vehicles. Consumer acceptance of this technology is critical to the
success of fully realizing the lifesaving potential of AVs. Trial by
error on public roads and risking public safety is neither the
appropriate nor the responsible approach to encouraging the development
and deployment of AVs. In fact, a national survey commissioned by
Kelley Blue Book found that a large portion of the public is resistant
to accepting AVs. Fifty-one percent of respondents replied that they
prefer to have full control of their vehicle, even if it's not as safe
for other drivers. Additionally, awareness of the higher levels of
vehicle autonomy is limited, with 6 out of 10 people saying they know
little or nothing about AVs. For half of the respondents, the
perception of safety and personal comfort with autonomous technology
diminished as the level of autonomy increased. In fact, 80 percent
believed that people should always have the option to drive themselves,
and nearly one in three respondents said they would never buy a level 5
vehicle.xxiii
---------------------------------------------------------------------------
\xxiii\ Kelly Blue Book, Future Autonomous Vehicle Driver Study
(Sept. 2016).
---------------------------------------------------------------------------
A recent study conducted by the Massachusetts Institute of
Technology garnered similar results. Only 13 percent of those polled
reported that they would be comfortable with vehicle ``features that
completely relieve the driver of all control for the entire drive.''
xxiv In addition, 59 percent of respondents reported that
the maximum level of automation that they would be comfortable with
were ``features that actively help the driver, while the driver remains
in control.'' xxv The reluctance and hesitation of the
public to embrace AVs will not be abated if the technology fails, the
government regulators are viewed as standing on the sidelines and
manufacturers have not done due diligence in conducting adequate
testing before rushing to the marketplace. Consumers expect, and in
fact, demand that the Federal Government prohibit the introduction of
dangerous products to the marketplace whether it be drugs, food, toys,
or driverless cars.
---------------------------------------------------------------------------
\xxiv\ H. Abraham, B. Reimer, B. Seppelt, C. Fitzgerald, B. Mehler
& J. Coughlin, Consumer Interest in Automation: Preliminary
Observations Exploring a Year's Change, Massachusetts Institute of
Technology, AgeLab, White Paper (2017-2), p. 6 (May 2017).
\xxv\ Id.
---------------------------------------------------------------------------
Recommendations:
All non-propriety communications and responses between the
agency and a manufacturer as it relates to any issues involving
AVs must be made available for public review and scholarly
research.
States Must Not be Preempted from Acting to Protect their Citizens
Especially in Light of NHTSA's Failure to Regulate Automated
Vehicles to Date
Advocates agrees with the statutory mission of NHTSA to regulate
the design and performance of motor vehicles to ensure public safety
which, in modern day terms, includes autonomous vehicles and
technology. However, unless and until NHTSA issues comprehensive
standards and regulations to govern the AV rules of the road, states
have every legal right, indeed a duty to their citizens, to fill the
regulatory vacuum with state developed proposals and solutions for
ensuring public safety. NHTSA, by issuing only guidelines, has left the
field of AV safety open to the states to fulfill their traditional role
of protecting the health and welfare of their citizens. As the National
Conference of State Legislatures (NCSL) noted in its comments to
NHTSA's guidelines, ``Without any indication on forthcoming Federal
regulations regarding the safe operation of HAVs, states may be forced
to fill the gap in order to ensure the safety of public
roadways.''xxvi Moreover, the Pennsylvania Department of
Transportation stated in its comments to the guidelines that ``Yes,
there should be consistent treatment of highly automated vehicles
nationwide. However, where the adoption of `safety standards' being
applied to highly automated vehicle testing is totally voluntary (as
opposed to self-certifying as against a regulatory framework in the
FMCSS) [sic], what level of comfort does that give to the states and
their citizens that their transportation and law enforcement agencies
are properly discharging their duty to ensure that highly automated
vehicles are in fact safe?'' xxvii
---------------------------------------------------------------------------
\xxvi\ William T. Pound, Executive Director, National Conference of
State Legislatures, Public Comments on Federal Automated Vehicles
Policy, Docket No.: NHTSA-2016-0090 (Nov. 21, 2016).
\xxvii\ Leslie R. Richards, Secretary of Transportation,
Commonwealth of Pennsylvania, Letter to Secretary Foxx and
Administrator Rosekind, Docket No.: NHTSA-2016-0090 (Nov. 21, 2016).
---------------------------------------------------------------------------
Recommendations:
Until NHTSA issues comprehensive standards and regulations
to govern the AVs, states must not be precluded from filling
the regulatory void with state developed solutions to protect
their citizens.
NHTSA Needs Additional Resources
The increase in motor vehicle fatalities combined with the demands
being placed on NHTSA with regards to the advent of AV technology
necessitates an increase to the agency budget. While the FAST Act did
provide some additional resources, it is still inadequate to manage the
myriad of challenges facing the agency. Today, 95 percent of
transportation-related fatalities, and 99 percent of transportation
injuries, involve motor vehicles on our streets and highways. Yet,
NHTSA receives only one percent of the overall U.S. Department of
Transportation (DOT) budget. NHTSA will face even greater challenges in
the future as AVs continue to develop and are introduced into the
market. For NHTSA to exercise proper oversight over AVs, the agency
will need to hire more staff with technical expertise. Moreover, given
that crashes impose a comprehensive cost to society of $836 billion,
$242 billion of which is direct economic costs such as lost
productivity, medical costs and property damage, it is critical to
advance serious measures to combat a serious problem.xxviii
---------------------------------------------------------------------------
\xxviii\ NHTSA Cost of Motor Vehicle Crashes Report.
---------------------------------------------------------------------------
Recommendation:
NHTSA must be given additional funding in order to meet
demands being placed on the agency with regard to the advent of
AV technology.
Conclusion
Autonomous vehicles have the potential to change our Nation's
ability to protect motorists and other road users. It is critically
important that we do everything possible to advance this life-saving
technology in as safe and expeditious manner possible. Advocates
believes that autonomous vehicles will not only change our lifestyles
but it may, once and for all, change our ability to achieve meaningful
and lasting reductions in the death and injury toll on our streets and
highways.
______
Summary of Statement of Jacqueline S. Gillan
Each year motor vehicle crashes kill tens of thousands and injure
millions more at a cost to society of over $800 billion. Advocates for
Highway and Auto Safety (Advocates) firmly believes that autonomous
vehicle (AV) technology has the potential to make significant and
lasting reductions in this mortality and morbidity toll.
The National Highway Traffic Safety Administration (NHTSA) has a
statutory duty to ensure the safety of AVs. Unfortunately, NHTSA has
chosen to issue only ``voluntary guidelines'' for the development of
AVs. Voluntary guidelines are not legally binding, are unenforceable,
and are inadequate to ensure safety and protect the public.
Vehicle and technology companies are already putting some AVs out
on public roads and beginning to market these systems. The development
of AV technology is not just taking place in the United States. In
fact, AVs are being tested throughout the globe in places such as the
United Arab Emirates and Singapore. Under current law, vehicles
imported from overseas have to meet the requirements of the Federal
Motor Vehicle Safety Standards (FMVSS). However, with only Federal
voluntary guidelines, foreign companies and entities may export
products that could be dangerously unsafe and NHTSA will have little
recourse. In addition, some experts predict that automated technology
will be placed in trucks before passenger vehicles. The potential for
an 80,000 pound truck using untested and unregulated technology on
public roads is a very real scenario if NHTSA relies on voluntary
guidelines and the ``good intentions'' of AV manufacturers.
A functional safety approach is essential to provide the framework
for the design, development and deployment of autonomous vehicle
technology. A similar process is currently utilized by the Federal
Aviation Administration (FAA) in portions of its regulations to ensure
safety while encouraging innovation.
Advocates agrees with the statutory mission of NHTSA to regulate
the design and performance of motor vehicles to ensure public safety
which, in modern day terms, includes AVs and technology. However,
unless and until NHTSA issues comprehensive standards and regulations
to govern the AV rules of the road, states have every legal right,
indeed a duty to their citizens, to fill the regulatory vacuum with
state developed proposals and solutions for ensuring public safety.
NHTSA's voluntary guidelines have left the field of AV safety open to
the states to fulfill their traditional role of protecting the health
and welfare of their citizens. Other commenters to NHTSA on voluntary
guidelines expressed a similar criticism about this regulatory vacuum
including the National Conference of State Legislatures (NCSL) and the
Pennsylvania Department of Transportation.
Advocates' Recommendations:
NHTSA must require that manufacturers meet a ``functional
safety standard'' to guarantee safety of AVs before they are
introduced into the marketplace.
Manufacturers must be required to submit a Safety Assessment
Letter (SAL) that confirms that the AV has been tested to
ensure it operates safely. The SAL should include all of the
test results and data to support the conclusion that the AV
system functions as designed.
NHTSA should be given the additional tools of imminent
hazard authority to protect against potentially catastrophic
defects with AVs and criminal penalties to ensure manufacturers
do not willfully mislead or misinform the agency.
Providing further broad statutory exemptions from the FMVSS
for AVs is both unnecessary and unwise. Until NHTSA revises the
FMVSS for level 4 and 5 AVs, manufacturers may seek appropriate
exemptions under the current process expanded by Congress in
the FAST Act (Pub. L. 114-94).
AVs must have adequate cybersecurity and privacy standards
to protect the public.
AVs should be subject to binding privacy standards to ensure
that consumer data is not used for purposes other than
improving the safety and security of AVs.
All non-propriety communications and responses between the
agency and a manufacturer as it relates to any issues involving
AVs must be made available for public review and scholarly
research.
Until NHTSA issues comprehensive standards and regulations
to govern the AVs, states must not be precluded from filling
the regulatory void with state developed solutions to protect
their citizens.
NHTSA must be given additional funding in order to meet
demands being placed on the agency with regard to the advent of
AV technology.
______
Advocates for Highway and Auto Safety
December 2, 2016
DOT Docket No. NHTSA-2016-0090
Docket Management Facility
U.S. Department of Transportation
Washington, DC.
Request for Comment on ``Federal Automated Vehicles Policy''
Notice and Request for Comments
81 Federal Register 65703, September 23, 2016
Advocates for Highway and Auto Safety (Advocates) files these
comments in response to the National Highway and Traffic Safety
Administration (NHTSA) notice and request for public comment on the
``Federal Automated Vehicle Policy.'' \1\ (NHTSA AV Policy). Simply
because AV technology has the potential to evolve rapidly over time
does not justify NHTSA abdicating its statutory mission of regulating
motor vehicles to ensure public safety.\2\
---------------------------------------------------------------------------
\1\ Request for Comment on ``Federal Automated Vehicles Policy,''
81 FR 65703 (Sept. 23, 2016).
\2\ In general the term AV refers to all autonomous vehicles
including those vehicles that NHTSA refers to as highly autonomous
vehicles (HAVs) except as noted.
---------------------------------------------------------------------------
Fifty years ago, Congress passed the National Traffic and Motor
Vehicle Safety Act of 1966 because of concerns about the death and
injury toll on our highways. The law required the Federal Government to
establish Federal motor vehicle safety standards (FMVSS) to protect the
public against ``unreasonable risk of accidents occurring as a result
of the design, construction or performance of motor vehicles.'' \3\
While cars have changed dramatically and will continue to do so in the
future, the underlying premise of this prescient law and the NHTSA's
safety mission has not.
---------------------------------------------------------------------------
\3\ Title 49, U.S.C. Sec. 30102.
---------------------------------------------------------------------------
Advocates has always enthusiastically championed technology and for
good reason. It is one of the most effective strategies for reducing
deaths and injuries. NHTSA has estimated that since 1960, hundreds of
thousands of lives have been saved by motor vehicle safety
technologies.\4\
---------------------------------------------------------------------------
\4\ Lives Saved by Vehicle Safety Technologies and Associated
Federal Motor Vehicle Safety Standards, 1960 to 2012, DOT HS 812 069
(NHTSA, 2015); See also, NHTSA AV Policy, Executive Summary, p. 5
endnote 1.
---------------------------------------------------------------------------
In 1991, Advocates led the coalition that succeeded in having the
airbag mandate included in the Intermodal Surface Transportation
Efficiency Act (ISTEA) of 1991.\5\ As a result, by 1997, every new car
sold in the United States was equipped with a front seat airbag and the
lives saved have been significant. In fact, airbags save over 2,000
lives annually.\6\ Advocates continued to build on our success by
pushing lifesaving technologies in other bills and regulatory
proposals. These efforts included tire pressure monitoring systems,\7\
rear outboard 3-point seat belts,\8\ electronic stability control,\9\
seat belt reminder systems,\10\ rear video cameras,\11\ brake
transmission interlock,\12\ seat belts on motorcoaches,\13\ electronic
logging devices \14\ as well as other important safety improvements
such as rollover crash avoidance and automatic emergency braking. These
safety advances have saved countless lives.
---------------------------------------------------------------------------
\5\ Pub. L. 102-240 (Dec. 18, 1991).
\6\ National Center for Statistics and Analysis, Lives Saved in
2015 by Restraint Use and Minimum-Drinking-Age Laws, National Highway
Traffic Safety Administration, Report No. DOT HS 812 319 (Aug. 2016).
\7\ Transportation Recall Enhancement, Accountability, and
Documentation (TREAD) Act, Pub. L. 106-414 (Nov. 1, 2000)
\8\ Anton's Law, Pub. L. 107-318 (Dec. 4, 2002).
\9\ Safe, Accountable, Flexible, Efficient Transportation Equity
Act: A Legacy for Users (SAFETEA-LU), Pub. L. 109-59 (Aug. 10, 2005)
\10\ Id.
\11\ Cameron Gulbransen Kids Transportation Safety Act of 2007,
Pub. L. 110-189 (Feb. 28, 2008).
\12\ Id.
\13\ Moving Ahead for Progress in the 21st Century (MAP-21) Act,
Pub. L. 112-141 (Jan. 3, 2012).
\14\ Id.
---------------------------------------------------------------------------
According to the latest statistics from the National Highway
Traffic Safety Administration (NHTSA), 35,092 people were killed on our
Nation's roads in 2015.\15\ This represents a 7.2-percent increase from
2014 and is the largest percentage increase in nearly fifty years.\16\
Injuries resulting from crashes also increased to 2.44 million from
2.34 million in 2014.\17\ Advocates is hopeful that automated vehicle
technology has the potential to significantly reduce this carnage.
However, the safety benefits of AVs will be realized gradually as the
widespread adoption of the technology will take years.
---------------------------------------------------------------------------
\15\ National Center for Statistics and Analysis, 2015 motor
vehicle crashes: Overview, Report No. DOT HS 812 318, National Highway
Traffic Safety Administration (Aug. 2016).
\16\ Id.
\17\ Id.
---------------------------------------------------------------------------
Introduction/Overview\18\
---------------------------------------------------------------------------
\18\ Due to the complexity of the issues involved and the length of
the NHTSA AV policy, Advocates' comments exceed the 15 page limit
pursuant to Title 49, C.F.R. Sec. 553.21.
---------------------------------------------------------------------------
The NHTSA AV Policy points out that under current law and
regulation, vehicle and equipment manufacturers are under no legal duty
to provide information to the agency about a new technology, in advance
of production and sale into the U.S. market, unless it fails to comply
with an applicable FMVSS or raises a compliance question regarding
existing regulations.\19\ The NHTSA AV Policy does nothing to change
the legal responsibility or duty that vehicle and equipment
manufacturers owe to NHTSA or the public. The NHTSA has the authority,
however, to require motor vehicle manufacturers, and other entities
supplying auto equipment, parts and electronic systems for AVs, to
conduct tests and perform analyses to document and verify that an AV
system performs safely and as designed. That type of documentation
should be required by rule rather than as part of the voluntary safety
assessment letter that the agency now requests manufacturers to
voluntarily submit to the agency.
---------------------------------------------------------------------------
\19\ Federal Automated Vehicles Policy: Accelerating the Next
Revolution in Roadway Safety, 12507-091216-v9, p. 48 (NHTSA, Sept.
2015) (NHTSA AV Policy), available at http://www
.nhtsa.gov/nhtsa/av/index.html.
---------------------------------------------------------------------------
Furthermore, the development and deployment of automated vehicles
as well as the agency's role in regulating this technology must be open
and transparent. Therefore, all communications and responses between
NHTSA and a manufacturer as it relates to any issues involving
automated vehicles must be made available for public review and
scholarly research. In addition, all data generated from the testing
and deployment of AVs, except for trade secrets and private individual
information must be made public. In the past few years, automakers have
hidden from regulators and the American public safety defects that have
led to unnecessary deaths and injuries as well as the recall of
millions of vehicles. This troubling lack of transparency should not be
allowed to infect the development of driverless vehicles. Lack of
transparency will severely undermine the public's confidence in this
new technology and inhibit its adoption by the public regardless of its
perceived benefits.
I. Vehicle Performance Guidance for Automated Vehicles
The Guidance
Advocates' two main objections to the proposed performance guidance
are that the guidance is not mandatory and that there is a lack of
specificity. In its current state, the non-mandatory guidance is such
that any information willingly provided by industry in the safety
assessment letters regarding compliance with the guidance may be
incomplete, sparse, or at best so varied from letter to letter as to
render the information collected useless in terms of agency review and
developing future regulation. As detailed below, the guidance should be
reconfigured around the concept of a functional safety approach to the
design, development, and deployment of autonomous vehicles of all
levels.\20\ The guidance must be specific in terms of minimal reporting
requirements and should establish those requirements based on a planned
path toward future regulation. Failure to do this will only leave the
Federal agency charged with ensuring motor vehicle and public safety
further behind the technology.
---------------------------------------------------------------------------
\20\ Levels in this case refer to the levels of automation as
defined by SAE J3016 and adopted by the NHTSA. Although the NHTSA
distinguishes between SAE Level 0-2 and 3-5 AVs, the agency
acknowledges that ``this distinction does not change many of the areas
in which the manufacturers . . . should apply elements of this Guidance
during product development, testing, and deployment.'' NHTSA AV Policy,
p. 31.
---------------------------------------------------------------------------
Scope
As indicated ``all individuals and companies manufacturing,
designing, testing, and/or planning to sell automated vehicle systems
in the United States'' \21\ must consider the guidance. Similarly, the
guidance must be applied to all AVs, including light-, medium-, and
heavy-duty vehicles. The guidance must apply to both test-and
production-level vehicles, specifically those which are sharing the
road with the general public.\22\ The reality is that while testing of
AVs on public roads is a necessity, to increase real-world data
collection and improve development, AVs are sharing the road with other
highway users who have not been informed of the testing, are unwitting
participants to the testing, and may be exposed to crash risks without
prior informed consent. AVs that are being tested as well as the routes
they will travel should be conspicuously marked so that they are easily
identifiable to the public. For these reasons, AVs used for research
and testing on public roads, as well as AVs sold to the public, must
all be subject to a rigorous functional safety process and requirements
to ensure that the public will not be exposed to an unreasonable crash
risk.
---------------------------------------------------------------------------
\21\ NHTSA AV Policy, p. 11. These comments use the term
``manufacturers'' to represent all entities intended to be covered by
the scope of the NHTSA AV Policy.
\22\ The public must be given the opportunity to give meaningful
informed consent to the testing of AVs on public roads. Among other
things, AVs that are being tested as well as the routes they will
travel should be conspicuously marked so that they are easily
identifiable to the public. In addition, as required by Federal
regulation for research involving human subjects, cities and states
that permit testing of AVs should have in place an Institutional Review
Board to monitor that the testing of AVs to protect the rights all of
the subjects involved in the testing. See: 45 CFR 46 (2009).
---------------------------------------------------------------------------
Overview: DOT's Vehicle Performance Guidance
In terms of content, the NHTSA AV Policy covers many of the
important aspects necessary to achieve safety of AVs. However, the
organization of the guidance appears disjointed and should be revised
to conform more closely to a functional safety approach. In its present
format, the guidance could be read as indicating that the individual
guidance sections could be tackled as separate, independent issues
when, in reality, a comprehensive and cohesive systems engineering
approach must be taken in order to achieve the safe deployment of AVs.
Almost all portions of the guidance represent, to some degree,
aspects of functional safety. Functional safety is a process by which a
system is designed, developed and deployed to ensure that the system,
as a whole, operates correctly and safely in response to inputs,
errors, and failures. Functional safety is applied throughout the life-
cycle of a system, from hazard analysis during design through auditing
of performance after deployment. Only through ubiquitous adoption of a
functional safety approach to the development of AVs can the safety and
benefits of this technology be achieved. Advocates recommends that
NHTSA reorganize the guidance into a framework focusing on a functional
safety approach that identifies how each of the guidance components fit
into the functional safety framework. The following is an example of
how different sections of the guidance could be reorganized into a
functional safety approach:
System Design
Operational Design Domain \23\
---------------------------------------------------------------------------
\23\ Operational design domain (ODD) refers to how the AV system
will detect and respond to the driving environment.
Object and Event Detection and Response \24\
---------------------------------------------------------------------------
\24\ Object and event detection and response (OEDR) refers to how
the AV system will perform when a problem with the system itself is
encountered.
---------------------------------------------------------------------------
Hazard Analysis
Mechanical
User
Human Machine Interface (HMI)
Consumer Education and Training
External
Cybersecurity
Federal, State, Local Laws
Ethical Considerations
Risk Reduction
Post-Crash Behavior
Fall Back (Minimal Risk Condition)
Design Validation
Validation Methods (simulation, track, on-road
testing)
Performance Verification/Auditing
Data Recording and Sharing
Safety Assessment Letter to NHTSA
The NHTSA must make the reporting of AV information mandatory.
Voluntary submission of information will not succeed because AV
manufacturers are under no legal duty to report completely and fairly.
While Advocates agrees with the underlying concept of the proposed
Safety Assessment submission, there are three major problems with the
Safety Assessment guidance as proposed in the NHTSA AV Policy. First,
at the outset, the NHTSA undermines its effort by describing the Safety
Assessment information to be provided as merely ``outlining'' how the
manufacturer submitting the information is meeting the areas of concern
in the NHTSA AVPolicy guidance.\25\ The agency then refers to the
Safety Assessment submission as a ``summary letter.'' \26\ In fact, the
agency guidance allows manufacturers to merely check-off a box for each
area of requested information that indicates whether the manufacturer's
AV system ``Meets'', ``Does not meet'', or ``is not applicable'' to
each particular guidance area of the NHTSA AV Policy. In essence, if a
manufacturer voluntarily responds at all, it could just check the
appropriate response in each area without providing any substantive
information or content whatsoever. Although the agency states that it
expects responses to be ``concise and complete[,]'' \27\ nothing in the
guidance indicates that the agency is seeking detailed information in
an initial response.
---------------------------------------------------------------------------
\25\ NHTSA AV Policy, p. 15.
\26\ NHTSA AV Policy, p. 16
\27\ NHTSA AV Policy, p. 16.
---------------------------------------------------------------------------
Second, as proposed, the request for the Safety Assessment
submission lacks specificity as to what type of information the agency
wants and that the manufacturer should submit. Requests for information
contain only vague descriptions that may or may not receive accurate
and complete responses. The agency approach to the Safety Assessment
submission letter is to allow the manufacturer to provide as much or as
little information as the manufacturer deems to be in its self-
interest. In fact, the agency states that, after the initial
submission, the agency ``might request more detailed information on
Guidance areas to better assess safety aspects of the HAV systems.''
\28\ It is critical that the agency should try to obtain complete and
detailed information from the outset. In its present non-mandatory
form, the agency will have little recourse to compel manufacturers to
provide additional information if the agency is not satisfied with the
initial response. Follow-up requests may well extend beyond the four-
month lead-time that the NHTSA AV Policy suggests is needed to review
manufacturer submissions prior to testing on public roads.\29\ In
addition, such requests impose additional burdens on NHTSA's resources.
---------------------------------------------------------------------------
\28\ NHTSA AV Policy, p. 16.
\29\ NHTSA AV Policy, p. 16.
---------------------------------------------------------------------------
Third, the NHTSA does not seek any uniformity of the substantive
content for the Safety Assessment information. Responses from
manufacturers can take many forms and use distinctive nomenclature that
will slow down if not stymie the agency's evaluation of the
information. The agency should categorize the types of information it
seeks in each guidance area in order to better be able to compare
implementations of industry standards, guidance, best practices,
testing, protocols, and analyses.\30\ Doing so will make the
information gathered more useful to the public, industry, and
government alike.
---------------------------------------------------------------------------
\30\ The lack of specificity with the Early Warning Reporting (EWR)
perfectly illustrates what can happen when the agency fails to provide
proper specification for the information it seeks. The vague categories
of the EWR have enabled the industry to hide questionable performance
and dangerous defects. (EWR: Elective Warning Reports--When
Manufacturers Don't Report Claims, Safety Research & Strategies, Inc.,
Apr. 8, 2014, available at http://www.safety
research.net/blog/articles/ewr-elective-warning-reports-when-
manufacturers-dont-report-claims)
---------------------------------------------------------------------------
Finally, regarding the timeline for Safety Assessment responses,
four months is excessive to develop a Safety Assessment letter for
products already on the road, and for which all documentation should
readily be available. This is particularly important as the initial
response may not be sufficiently detailed and may require the agency to
make an additional follow up request for more information. Likewise the
demand that letters be provided four months prior to testing should be
evaluated to ensure that the agency and the public have ample time to
review the documents, especially if a vehicle is to be tested on public
roads. For similar reasons, Advocates concurs that updates to the
Safety Assessments are needed, and should be required, when updates are
made to AV systems. However, the process and requirements for updating
Safety Assessments should be specific and uniform to ensure that the
information gathered is beneficial to the public, industry and
regulators.
Cross-Cutting Areas of Guidance
Data Recording and Sharing
The collection and sharing of data with the public, the NHTSA, and
within the industry will be critical to achieving the highest levels of
safe AV performance. Similar to the way in which current crash
databases are used to inform current safety regulation and vehicle
design, an AV performance database would benefit all parties. NHTSA
should maintain a public database that details any and all crashes
involving an AV. The sharing of both incident (crash) and near incident
(near-miss) data would enable the industry to review performance of
other AV systems and improve their own performance, especially for edge
cases.\31\ Likewise, the agency should use the data to develop specific
sets of scenarios which AV systems must be able to address as part of
future guidance and eventual regulation. The data to be collected and
shared must be more specific than ``all information relevant to the
event and performance of the system.'' \32\ This vague description
leaves too much room for interpretation which could result in
disjointed and unusable information from which limited insights could
be gained. Importantly, specification of data to be recorded and shared
should address industry concerns around intellectual property. For
example, making sure that the data from an incident where an AV fails
to see the side of a white truck crossing its path is of critical
importance to the development of all AV systems as it identifies a
scenario which could cause problems and lead to risks. This sharing of
data or feedback loop, must be done to ensure that each AV system does
not have to learn only from its individual failures but can be improved
based on the failures or successes experienced by other AV systems.
---------------------------------------------------------------------------
\31\ Edge Cases are those problems or situations that occur very
infrequently or at the operational boundary. These are cases each
manufacturer's vehicles may only see once, but the information about
that situation would benefit all systems because of its rarity.
\32\ NHTSA AV Policy, p. 18.
---------------------------------------------------------------------------
Essential to gathering accurate and reliable technical data on AV
system performance and failure is the need to require all such data to
be captured and collected. While each manufacturer may voluntarily
provide information that is collected for internal use, information
about AV system performance on public roads must be provided to the
NHTSA on a real time basis and made available to the public. In the
meantime, the NHTSA must also require that Event Data Recorders (EDR)
or other systems are able to collect and record all essential data on
AV systems so that in the event of a crash or other failure the vehicle
systems status data will be available to the agency, crash
investigators, researchers and the vehicle owner. Advocates recommends
that the agency complete the rulemaking to mandate EDRs in all
passenger vehicles, and revise the current requirements in the EDR
rule, 49 C.F.R. Sec. 563, to require additional data collection on all
AV systems.
Privacy
Privacy will be a key factor in ensuring acceptance of AVs by the
public. As noted above, the recording and sharing of data will be
critical to achieving the safest performance of AVs as quickly as
possible. Keeping the public informed as to the importance of sharing
data and ensuring their privacy will be similarly critical to ensure
participation. The industry and regulators must guarantee that data is
protected and only used for the purposes of improving safety, and not
for other commercial uses which could turn the public against data
sharing.
System Safety
Functional safety should provide the framework for the overall
approach to the guidance and future regulation. In almost all areas of
the guidance, it will benefit all parties involved if the requirements
of the safety assessment were more specific in terms of standards,
guidance, best practices, and design principles about which the agency
would like information. The industry could then respond as to whether
and how they have implemented any of those practices. Requiring the
specific and uniform reporting of data by manufacturers will enable the
agency to understand how each of the guidance areas, practices,
standards, etc., are being implemented across the industry. It will
also allow the agency to work towards establishing regulation to
require their implementation.
Vehicle Cybersecurity
Cybersecurity is an important aspect of AV development which must
be addressed as part of functional safety. Again, the guidance is vague
on the information it would like to receive from the industry.
Specificity and uniformity will be the keys to evaluation of AV system
information by the NHTSA and to informing the public. The agency should
identify problems areas and require specific responses from
manufacturers on how those are being addressed. Problem areas could
include subjects such as GPS signal loss/degradation/spoofing, and off-
line and real time hacking of single vehicles or fleets of vehicles. As
with all other performance data, the sharing of data in terms of
cybersecurity will improve overall safety and ensure that all vehicles,
regardless of manufacturer, are afforded the same level of security.
This is even more important when the aspect of connected vehicles is
concerned as any weak link in the chain could leave many more vehicles
vulnerable to programming errors or hacking. The potential risk of a
single software error, or malevolent computer hack impacting hundreds
or thousands of AVs, perhaps whole model runs, makes appropriate
cybersecurity a crucial and indispensable element of AV design.
Human Machine Interface
The user interface is an essential aspect of the development and
deployment of AVs. In the functional safety approach, the human/machine
interface (HMI) presents both a source for and means of addressing
hazards stemming from the user. As some products currently on the road
have demonstrated, poor HMI design can lead to dangerous and deadly
situations. For example, if an AV requires a human occupant to
participate in the driving process, it must be designed to ensure that
the human occupant is engaged, aware, and informed of the operational
status of the vehicle particularly in time critical settings. Simply
informing drivers that they must remain engaged or placing information
in the owner's manual as to the limitations of the AV system are not a
sufficient or acceptable substitute for engineering solutions that are
effective in maintaining the engagement of the human operator. This is
true even if the operator signs a statement indicating that they have
read the manual. In a functional safety approach the hazard of driver
engagement should be managed through monitoring and warning systems to
ensure drivers are engaged to the level necessary for the system to
operate safely. A system which fails to account for all sources of risk
and hazard, especially the transition from or to a human operator,
would not be functionally safe and should be evaluated by the agency
before the AV system is certified by the manufacturer for use on public
roads.
Crashworthiness
Compliance with the FMVSS is mandatory and should remain so. In the
future, as new seating configurations are proposed, the NHTSA may be
required to adapt the FMVSS to ensure that occupants are provided the
same minimal levels of crashworthiness protection currently afforded by
all vehicles. Leaving occupant safety in AVs to the ``due care'' of the
industry is a step backward that is fraught with problems and opens the
door for making trade-offs with safety. Furthermore, concerning AV
compatibility, the agency should establish regulatory requirements
rather than relying on voluntary agreements. In the end, it is
foreseeable that AVs will share the road with traditionally operated
vehicles with human drivers for an extended time and, despite any AV
advances, will likely still be involved in crashes which will require
the vehicle design to ensure that it protects occupants and crash
partners alike.
Consumer Education and Training
Consumer education and training are imperative to ensure safe
deployment of AVs. Failure to fully familiarize consumers with needed
operational and safety information, or to properly train vehicle owners
who may have to take over operation of the AV at some point, could
result in rejection of AVs by the public and more importantly lead to
crashes. As end users of the product, even if they are not involved in
the driving task at all, consumers will ultimately decide the fate of
AVs. Consumer education and training in all aspects of AV operation are
critical to success.
Advocates agrees with the NHTSA's statements and observations in
the guidance that manufacturers develop education and training programs
for employees, dealers, distributors and consumers.\33\ This is
especially necessary for Level 3 (and lower) AVs in which the driver
may need to take over control of the vehicle when the operational
limits of the AV system have been reached. The suggestion in the NHTSA
AV Policy that consumers who purchase AVs should receive training on
the operational capabilities and limitations of the AV system they
purchase, along with an on-the-road demonstration prior to taking
possession of the AV would be beneficial. However, the agency should
take the next step and offer specific solutions and develop prototype
program materials to guide manufacturers in this effort.
---------------------------------------------------------------------------
\33\ Advocates is aware of instances in which dealership sales
personnel were unfamiliar with the capabilities of the AV systems they
were demonstrating to customers.
---------------------------------------------------------------------------
The agency must also consider the need for standardizing the
operation of AVs or their functions, and requiring training and
education/information from the manufacturers or as part of the state
policy for licensing to ensure that, until such time as drivers are no
longer required, that lack of education or training do not undermine
the safety benefits of AVs.
Furthermore, while all relevant information must be included in the
vehicle owners' manual, there is a clear need, especially for vehicles
that may require the intervention of a driver or which can have their
operational capabilities updated significantly, for important features
of the AV system operation to be delivered to consumers in other ways.
The agency should consider requiring instrument panel instructions/
notifications that must be read and accepted before the AV can begin
operating, and a separate AV system tutorial that can be viewed on
board the AV when not in operation or on other personal data devices. A
comprehensive approach to AV education and training is essential for
the success of AV adoption.
Certification
Manufacturers will still be responsible to self-certify that their
AV meets all Federal and state requirements as well as operates safely
and as designed. Manufacturers and other responsible entities will need
to keep both the NHTSA and consumers aware of the operation,
capabilities, and limitations of AV systems that are on the road. With
the possibility of over the air (wireless) updates which could change
the operation of AVs overnight, it will be essential for manufacturers
to provide updated Safety Assessments to the NHTSA regarding any
changes that affect the certification of the vehicle. Clear
instructions regarding any changes must also be conveyed to the vehicle
owner and operators.
Post-Crash Behavior
As part of functional safety, AVs must be able to recognize
failures and address them or enter a failsafe mode. Controlling post-
crash behavior to prevent the operation of AVs systems after damage to
sensors is part of addressing a crash which is a known hazard.
Similarly, it is important for the repair and re-certification of AV
systems to be well established to ensure only safe AV systems are
allowed back onto public roads.
Federal, State and Local Laws
In terms of functional safety, complying with Federal, State, and
local laws is a known aspect of an AV systems design and expected
operation. Variation in laws between jurisdictions is another known
operational parameter that must be considered and addressed by AV
manufacturers. It would appear that compliance with Federal, State, and
local laws would be part of the operational design domain (ODD) and the
object and event detection and response (OEDR), rather than a stand-
alone topic of the guidance. Additionally, the NHTSA should consider
the impact of the variation in transportation laws, road designs, lane
marking, signage, etc. on the implementation of AVs and the benefits
which could be derived from seeing uniformity across the country.
Ethical Considerations
Conflicts between the objectives of safety, mobility, and legality
could occur in the operation of AVs; however, the guidance is vague on
the specific means by which these conflicts will be resolved. Merely
stating that solutions to these issues ``should be developed
transparently using input from Federal and State regulators, drivers,
passengers and vulnerable road users . . .'' \34\ is not sufficient to
ensure that dangerous products are not placed on the road now. Until
such time as these ``algorithms'' are developed and proven, the safety
objective should guide decisions. It is for this reason that the
collection and sharing of data on vehicle operation will be important.
The NHTSA must be more forthcoming with regard to how it will approach
AV designs in which the manufacturers have opted to make different
decisions in balancing these ethical considerations.
---------------------------------------------------------------------------
\34\ NHTSA AV Policy, p. 26.
---------------------------------------------------------------------------
Automation Function
Specification of the conditions under which an AV system will
operate (ODD), how the AV system will detect and respond to the driving
environment (OEDR), and how the AV system will perform when a problem
with the system itself is encountered (fall back (minimal risk
conditions)), are all part of the design aspect of a functional safety
approach. For each system and the system overall, specifying which
conditions a system can operate within, what the expected response is,
and what happens when all else fails are essential parts of the design,
hazard analysis, and hazard mitigation/elimination aspects of
functional safety. The NHTSA must ensure that it collects sufficient
information and test data/results to confirm that the manufacturer has
done its safety due diligence and to validate that these aspects of the
AV system are operating safely and as designed.
Operational Design Domain
Defining an ODD and translating that information to the NHTSA and
specifically the consumer will be important to ensure that user error
is reduced and that limitations in the operation of AV systems both
within and outside of the ODD can be identified. Specificity and
uniformity in reporting will improve the ability for this information
to be used to develop future regulation and inform the public. It is
also important that manufacturers consider specifying not only the ODD
(where the system will work) but to clearly define the operational
boundaries for the consumer. As discussed in previous sections, simply
informing the consumer is not adequate from a functional safety stand
point and should be supplemented with engineering solutions, including
properly designed HMI, to ensure that opportunities for error or
misunderstanding that could pose a crash risk are eliminated.
Object and Event Detection and Response
Within its ODD, each AV system must be expected to detect and
respond to all scenarios which could affect safe operation. This
includes interactions with other vehicles, pedestrians, cyclists,
animals, and other objects. The NHTSA provides a limited list of
behavioral competencies for normal driving and references a number of
scenarios for crash avoidance and other hazards (construction, disabled
vehicles, etc.). Advocates believes that this type of specific listing
of scenarios which all AVs, as appropriate based on each particular AV
system ODD, should be addressed as part of the safety assessment. With
sufficient, uniform data sharing, manufacturers should be aware of all
scenarios that AVs should address as part of their safety assessment.
The agency should work towards a functional safety requirement which
would include specifications of scenarios which must be addressed by an
AV system, depending on the ODD, and which could be tested to ensure
compliance.
Fall Back (Minimal Risk Condition)
As noted above in the discussion of post-crash behavior and in
other sections, establishing fail safe conditions and operation is yet
another part of functional safety. Once hazards and risks are
identified and then eliminated, mitigated, or guarded against, and
warnings are provided, the final step is to ensure that a system can
recognize a failure or when it is operating outside of its ODD and
return the system to a failsafe operation. However, having a failsafe
design should not just be a recommendation. As with nearly all other
parts of the guidance, AVs must be functionally safe, including having
failsafe modes. These modes will be especially important in the early
rollout of AVs when unknown risks have not yet been adequately
identified through data collection and sharing.
Validation Methods
Performance validation is another step in the functional safety
process. The NHTSA Safety Assessment must include specific tests and
validation methods which the AV manufacturers must confirm have or have
not been used, with a description and documentation of the methods the
manufacturer did use to validate its AV system. Manufacturers should be
required to provide information on all methods beyond those listed by
the agency which will inform and enable the agency so it can update
future versions of the list. Again, data collection and sharing will
also be critical to performance validation to ensure that performance
on the road and in the hand of the public is matching the performance
predicted by the design.
Guidance for Lower Levels of Automated Vehicle Systems
All manufacturers of AV systems should be required to meet
functional safety requirements for the design, development, and
deployment of AVs. Almost all of the current guidance fits into what
should be required in a comprehensive functional safety approach. Lower
levels of automation should not be exempt from having to thoroughly
apply the process.
Advocates supports ODD, OEDR, and minimal risk conditions applying
to lower level AV systems. Every AV system has conditions under which
it is designed to operate and its operation is specified. While the
details may not be as extensive as that of a higher level (3-5) system,
this is not an exception but rather a modification. Finally, a failsafe
mode is still necessary even if it is not as drastic as those for
higher level AVs. Uniformity in application of the functional safety
process across all levels of AVs is necessary to ensure safety and
inform the agency and the public.
Next Steps: Activities to Improve, Expand and Oversee the Guidance
It is critical that the NHTSA move swiftly towards effective
regulation before the technology becomes ubiquitous. Abandoning the
regulatory process is not a solution. A concerted and coordinated
effort between agencies, branches of government, industry and the
public will be necessary to achieve the goals of bringing AVs to market
and doing so safely.
II. Model State Policy
The stated purpose of the model state policy is to ``create a
consistent, unified, national framework'' for AV regulation, yet the
framework is so loose it is difficult to envision consistent
outcomes.\35\ In fact, this framework, because it lacks so many
necessary details, will create a patchwork of state standards which
stands in stark contrast to the stated goals of the model policy. The
NHTSA fails to include a timeline for suggested evaluations and actions
that states need to complete. While the model policy acknowledges that
states must undertake changes and improvements to transportation
infrastructure in order to ensure the safe operation of AVs, it fails
to instruct the states on what specific upgrades are needed. In
addition, the model policy does not set any standard for crash
reporting deadlines or data requirements which will be crucial to state
regulation of the testing and deployment of AVs. Not only are the
paucity of details contained in the model policy of great concern, the
proposal also contains several critical shortcomings on its face that
must be rectified.
---------------------------------------------------------------------------
\35\ NHTSA AV Policy, p. 37.
---------------------------------------------------------------------------
Introduction
Advocates agrees with the statutory mission of NHTSA to regulate
the design and performance of motor vehicles including AV technology
and vehicles to ensure public safety. However, so long as the agency
chooses to issue merely voluntary guidance to address safety and
regulatory issues on emerging AV technology, states have every legal
right to fill the regulatory vacuum with state developed solutions.
NHTSA, by issuing only guidelines, has left open the field of AV safety
that can and must be filled by the states. As such, each state can
currently dictate (and some already are) what tests it deems acceptable
and what constitutes an acceptable or successful result. This
decentralized process leaves an incredible amount of variability in
standards and interpretation. Advocates urges the agency to issue
regulations to govern the safety assessment process and other
regulatory appropriate aspects of AVs that the agency has exercised
with respect to the safety of non-automated motor vehicles.
NHTSA should partner with states to properly train and educate
drivers on the proper use of AV systems; however, specific focus should
be placed on training dealership sales personnel and vehicle buyers and
drivers on the capabilities and limits of AVs including when a driver
must retake control of an AV after disengagement of the autonomous
system.
NHTSA should take a leadership role and work with the states to
ensure enhanced requirements for AVs. The inadequate maintenance of an
AV, particularly out of date software, could have disastrous
consequences on its operation and public safety. These concerns merit
that AVs receive a heightened standard of care from state regulators
tasked with ensuring that all vehicles registered in a state are
properly maintained.
Model State Policy
The NHTSA must require, and recommend that state law reinforce, the
need for AVs to be designed to comply with all aspects of state motor
vehicle and traffic laws. Furthermore, NHTSA should work with the
Federal Highway Administration, the Federal Motor Carrier Safety
Administration and the states to ensure that AVs are compliant with all
highway and traffic requirements. Mandating this uniformity will
greatly reduce the opportunity for errors by AVs when dealing with the
current patchwork of state requirements or guidelines such as lane
widths, road markings, and signage.
In terms of defining who the ``driver'' of an AV is the NHTSA AV
Policy contains an apparent inconsistency. For level 3 AV systems,
where the human driver has to be able to take over control of the
vehicle, states should define the ``driver'' of the vehicle as the
human operator of the vehicle not the AV system as proposed in the
model state policy. NHTSA asserts that States should generally deem the
AV system to be the ``driver'' of the vehicle for SAE Levels 3-5.\36\
Later in the guidance, however, the NHTSA indicates that States should
continue to regulate the ``human'' driver for AV ``technologies that
are less than fully automated, SAE levels 3 and lower.'' \37\ For a
level 3 vehicle where a driver is still tasked at various times with
the operation of the vehicle, when the vehicle reaches its operational
limits, the driver of such a vehicle must continue to be defined as the
human operator.
---------------------------------------------------------------------------
\36\ NHTSA AV Policy, p. 39.
\37\ NHTSA AV Policy, p. 43.
---------------------------------------------------------------------------
Administrative
Advocates supports the establishment of a jurisdictional automated
safety technology committee in each state as outlined in the model
policy. However, the committees should be balanced and include strong
representation from safety and consumer representatives as permanent
official members. Placing these representatives on the Committee will
foster public confidence in the testing and deployment of AVs in their
state and will encourage acceptance of the technology as it becomes
more prevalent. In addition, these representatives can provide a unique
and invaluable perspective to the Committee as they consider numerous
issues affecting consumers and public safety.
Jurisdiction Permission to Test
The lead state agency tasked with regulating AVs should prohibit
the testing of AVs in safety sensitive areas such as school and
construction zones as outlined in the model state policy. The
malfunction of an AV while testing in such areas would be catastrophic.
In addition, authorizations given for testing should be renewed on an
annual basis due to the number of issues that could evolve during the
course of a year involving experimental vehicles using such rapidly
developing technology that could receive and require programming
changes and updates at any time. Moreover, any authorization should
immediately be reviewed after a notice of a crash involving an AV
covered by the permit.
Testing by the Manufacturer or Other Entity
As outlined in the model state policy, not only should all crashes
involving test vehicles be reported to the applicable state regulatory
body, but any and all safety critical events such as near misses or
operational malfunctions should also be disclosed to state regulators
and the public. State regulatory bodies cannot accurately assess the
progress of AVs and their ability to operate on public roads without
this full and accurate picture of the performance of AVs during
testing.
Law Enforcement Considerations
Advocates concurs with NHTSA's statement that ``[f]or vehicles that
offer less than full automated capabilities, there is potential for
increased distracted driving.'' \38\ Despite these obvious concerns,
the agency continues to merely issue guidelines rather than regulations
to govern in-vehicle and nomadic electronic devices that contribute to
distracted driving, an issue that has long been identified by the
safety community and the agency and supported by numerous research
studies as a major public health epidemic.\39\ For AVs that are less
than fully automated, states must be encouraged to enact robust laws to
govern any and all in vehicle distractions.
---------------------------------------------------------------------------
\38\ NHTSA AV Policy, p. 45.
\39\ Distraction.gov, NHTSA.
---------------------------------------------------------------------------
III. NHTSA's Current Regulatory Tools
Rulemaking Authority
The NHTSA's most broad-ranging authority to influence and govern
motor vehicle safety, and to eliminate unreasonable risks to the
public, is through proposing a standard to govern conduct and
performance through a public rulemaking process. This is, and has been
since the inception of the agency, the approach used to ``adopt new
standards, modify existing standards, or repeal an existing standard.''
\40\ The establishment through regulation of safety standards for motor
vehicle safety and related technology performance requirements is the
standard and established agency procedure. The current body of
lifesaving Federal motor vehicle safety standards (FMVSS) has been
constituted through this process and procedure.
---------------------------------------------------------------------------
\40\ NHTSA AV Policy, p. 49.
---------------------------------------------------------------------------
Public notice-and-comment rulemaking provides a number of important
benefits. The regulated industry and the public receive a concise,
detailed statement of agency plans and proposed rules, including
testing procedures and performance requirements. It requires the agency
to master detailed technical issues and to organize the analysis of
benefits and costs of a proposed rule. The regulated entities and the
public are afforded an opportunity to evaluate and analyze the proposal
and provide focused feedback to the agency, including posing questions
regarding matters that may require greater clarity as well as critical
feedback on technical aspects of the rule. Rulemaking affords all
stakeholders, including the public and the regulated industry, the
platform for making broad, general philosophical statements regarding
the agency proposal, as well as the avenue to address narrow, technical
issues. The regulatory process clarifies the agency's intent as to what
will be subject to a particular regulation and how the agency intends
to regulate an aspect of motor vehicle equipment or operation. This is
important to reach a general agreed understanding of the issue and
assure only sound, well-reasoned requirements are imposed on the
regulated industry. While rulemaking may, by its nature, take longer
than other existing regulatory tools (e.g., interpretation letters or
exemption petitions), rulemaking according to the agency ``enables the
Agency to make the broadest and most thorough changes to governing
regulations, and gives the public the greatest opportunity to
participate in the Agency's decision-making process.'' \41\ In fact,
the NHTSA states that rulemaking may be the best approach to address
``a motor vehicle or equipment design [that is] substantially different
from anything currently on the road [for which] compliance with
standards may be very difficult or complicated . . .'' \42\ This
description directly applies to the development and installation of
AVs.
---------------------------------------------------------------------------
\41\ NHTSA AV Policy, p. 50.
\42\ NHTSA AV Policy, p. 49.
---------------------------------------------------------------------------
The future reliability and public acceptance of AVs would benefit
greatly from regulatory action that sets a fair and level playing field
for industry and, at the same time, provides transparency and oversight
for the introduction of AVs into the motor vehicle fleet. While the
NHTSA has not regulated every aspect of motor vehicles, crucial safety
and operating systems have been regulated as part of the FMVSS or other
pertinent regulations for decades. There is no clear and compelling
reason why AV technology should be treated any differently or given
greater leeway than previous mechanical or electronic technological
innovations. Furthermore, there is no reason to believe that AV
technology differs in any meaningful degree from the developments and
improvements that have been routinely regulated over the past 50 years
of automotive development. The agency cannot evade its statutory duty
simply because the new technology seems complicated or highly
technical.
The NHTSA has expressed its view that ``only after new technology
is developed and proven does the Agency establish new safety standards.
* * * Strong safety regulations and standards are a vital piece of
NHTSA's safety mission and the Agency will engage in rulemaking related
to automated safety technologies in the future.'' \43\ While Advocates
understands that, at present, there may not be sufficient information
and data to establish minimum performance requirements for some aspects
of AV systems, that does not mean that the agency should defer from
regulating those aspects of AV performance that are already known but
which are not yet regulated, such as autonomous emergency braking (AEB)
systems. Moreover, NHTSA is not precluded from requiring manufacturers
to adopt a functional safety process that will ensure appropriate
testing of AV systems will be conducted to prove the systems' safe
performance within the design parameters of the particular AV system.
---------------------------------------------------------------------------
\43\ NHTSA Enforcement Guidance Bulletin 2016-02: Safety-Related
Defects and Automated Safety Technologies, Final Notice, NHTSA, 81 FR
65705, 65706 (Sept. 23, 2016).
---------------------------------------------------------------------------
The NHTSA has an obligation to the public to ensure that new,
highly complex AV systems will perform safely. At this time the NHTSA
should require each vehicle manufacturer, or other company producing
the computer logic and software, to adopt a functional safety process
that requires comprehensive testing of AV systems that the agency can
review. This would ensure that the manufacturers are under a legal
obligation to perform appropriate due diligence to disclose the results
to the agency prior to motor vehicles equipped with AV system's being
placed in the stream of commerce.
The agency can and should require manufacturers, as part of the
self-certification requirement, to certify that sufficient functional
safety testing and analysis has been performed to establish that the AV
system will perform safely and as designed. To accomplish this, the
agency should require all AV system manufacturers, by regulation, to
adopt a functional safety process to conduct state-of-the-art testing
and analysis to establish that the AV system performs safely and meets
or exceeds all aspects of the systems design parameters. Such a
requirement is intended to allow the manufacturer to document that its
AV system(s) has been fully and extensively tested and that all
critical features, including the programming software, operate
properly. There must be a modicum of regulatory oversight by NHTSA to
ensure that the AV system manufacturers are acting responsibly and have
not taken any short-cuts in the rush to market.
Advocates' call for a regulation to require functional safety
testing and analysis is based on empirical evidence. The May 7, 2016
crash of a Tesla Model S equipped with the Autopilot AV system feature
is a real world example of why such regulatory action is necessary as
well as the timely notification to NHTSA is essential.\44\ Tesla has
stated that it required drivers to acknowledge that they would remain
engaged in the driving task. A functional safety analysis would have
required Tesla to evaluate what happens when the system fails to
identify a road hazard, when the driver is not engaged in the driving
task, or when both failures occur simultaneously all of which are
foreseeable. Functional safety requires consideration of all
foreseeable and feasible failure modes, as well as unlikely and low
probability failure modes. Blaming the driver for failing to heed a
warning, or for over reliance on the AV system, is inexcusable when
technological means of mitigating the risks of sensor failures and
human behavior are available.
---------------------------------------------------------------------------
\44\ The Tesla Autopilot System has been the subject of
investigation in other crashes, including another fatal crash in China
in January of 2016, Neal Boudette, Autopilot Cited in Death of Chinese
Tesla Driver, The New York Times (Sep. 14, 2016).
---------------------------------------------------------------------------
The risk classification for failures in this and similar systems
being developed must be very stringent since failure of the Autopilot
sensors or the detection algorithm while travelling would present an
unreasonable crash risk. Such an analysis, likely, would have led to
requirements that additional sensors be added to the vehicle to
mitigate the sensor ``blind spot'' created by using visual data from a
camera with limited input from additional technologies such as radar or
lidar. Tesla subsequently announced that it had fixed the problem
through a new software update to the Autopilot System.\45\ This flaw
would have and should have been detected by Tesla during a functional
safety analysis of the Autopilot System. Without such a required
analysis, there may be other instances where readily detectible and
foreseeable problems were not identified until after a crash occurs.
---------------------------------------------------------------------------
\45\ Louis Hansen, Tesla turns to radar in upgrade of Autopilot,
The Mercury News (Sept. 11, 2016). In fact, Tesla has not provided any
evidence that the problem has been fixed.
---------------------------------------------------------------------------
The functional safety analysis might also have resulted in
additional driver engagement requirements such as driver monitoring and
warning systems, and stronger driver reengagement methods to ensure
driver readiness to retake control of the vehicle. In the functional
safety analysis, Tesla would have been required to research, test, and
examine not just the Autopilot sensor array, but driver reengagement
readiness to prove that the design enabled drivers to either remain
engaged in the driving task (as Tesla stated was necessary) or that re-
engagement could be successfully accomplished in a safe manner. In
fact, after the fatal crash the company instituted more warnings to the
driver to remain alert while the Autopilot System is engaged.\46\ It is
the NHTSA's responsibility to require functional safety certification
to eliminate the types of system failures that resulted in a fatality
and eliminate the need for ``after the fact'' improvements which should
have been anticipated.
---------------------------------------------------------------------------
\46\ Id.
---------------------------------------------------------------------------
While the NHTSA's proposed voluntary submission of a letter
(addressed above) has many of the hallmarks of what is needed to be
reported to the agency, a regulation must, first and foremost, require
manufacturers of AV systems for motor vehicles to adopt a functional
safety process that will ensure that the manufacturer tests its AV
system appropriately and comprehensively. Second, documentation of all
test results, those that establish the functional safety of the system,
as well as any information that indicates the failure of the AV system
to fulfill functional safety parameters, must be provided to NHTSA in
advance of placing the AV system or AV equipped vehicle in the stream
of commerce. Finally, these aspects must be made mandatory, rather than
voluntary, in order to ensure that the manufacturers are legally
obligated to adopt an effective functional safety process and to supply
all relevant information to the agency.
Enforcement Authority
In relation to the NHTSA's enforcement authority, the NHTSA AV
Policy states that ``when vulnerabilities of [automotive] technology or
equipment pose an unreasonable risk to safety, those vulnerabilities
constitute a safety-related defect'' \47\ that can spur the agency to
investigate and take remedial action including ordering the
manufacturer to conduct a safety recall.\48\ Moreover, the agency's
enforcement authority applies `` `notwithstanding the presence or
absence of an FMVSS for any particular type of advanced technology.' ''
\49\ Advocates strongly supports the NHTSA's enforcement efforts to
investigate, identify and recall noncompliant or defective vehicles.
Advocates views the enforcement role played by the agency as critical
to ensuring that motor vehicles, once in use, perform safely. However,
the agency position begs the question as to why the agency would forego
issuing a prospective regulation to improve the safety performance of
production models and eliminate safety flaws in AV systems before they
are sold to the public. The NHTSA's enabling statute recognizes that
the agency must act in a proactive manner to avoid and reduce crashes,
injuries and deaths in the first place.\50\ Rather than relying
entirely on its enforcement authority, the agency should also be
actively ensuring through regulatory, as opposed to only voluntary,
action that AV systems are produced without production flaws before the
AV systems are sold to the public, placed in the stream of commerce,
and then subsequently have to be recalled.
---------------------------------------------------------------------------
\47\ NHTSA AV Policy, p. 50.
\48\ NHTSA AV Policy, p. 50.
\49\ NHTSA AV Policy, p. 50, citing NHTSA Enforcement Guidance
Bulletin 2016-02: Safety-Related Defects and Automated Safety
Technologies, 81 FR 65705, 65707 (September 23, 2016).
\50\ 49 U.S.C. Sec. 30101.
---------------------------------------------------------------------------
The historic and well-founded approach of the NHTSA has been to
require, in the first place, specific minimum performance requirements
or standards that vehicle manufacturers must certify that each vehicle
meets. These standards raise the quality and safety performance of
motor vehicles before sale to the public. While its enforcement
authority does permit the agency to investigate and recall vehicles for
safety defects even if there is no underlying performance standard, the
lack of an underlying performance standard may complicate the exercise
of the agency's enforcement authority. This is a significant concern
for electronically controlled systems and software where a specific
flaw or malfunction may be more difficult to identify. If there is no
underlying requirement or process, manufacturers will be far less
cautious about ensuring the operating safety of the vehicle prior to
sale.
Finally, although the NHTSA's enforcement authority is extensive,
safety flaws may take years to identify and recall. This has been the
situation with a number of safety recalls. In 2000, congressional
hearings and the media revealed hundreds of needless deaths and
injuries caused by the Firestone/Ford Explorer defective tire fiasco
because of delayed agency action. Again, for the same reason, in 2009,
families were put at unacceptable risk due to the Toyota sudden
acceleration problem. In the past several years, the public has learned
about the cover-ups and deception by General Motors which knowingly
used faulty ignition switches that have been linked to at least 169
deaths and many more injuries because of long delayed agency
action.\51\ Furthermore, the defective air bags manufactured by Takata
have resulted in millions of vehicle recalls and have caused at least
11 deaths in the U.S. and numerous injuries because, in part, of
delayed agency action.\52\
---------------------------------------------------------------------------
\51\ Angelo Young, General Motors Ignition Switch Flaw Death Toll
Rises To 169 Following Personal Injury Settlement; Penalty Costs Top
$2.1B, International Business Times (Sept. 17, 2015)
\52\ Ryan Beene, 11th U.S. death related to Takata airbags reported
by Honda, Automotive News (Oct. 20, 2016).
---------------------------------------------------------------------------
Although its enforcement powers are necessary, they are not
executed immediately after a safety defect involved crash occurs. It
may then take months or years to establish that a defect exists and
what the defect is and then more time to convince the manufacturer that
a recall is appropriate and, even then, further negotiation to ensure
that the scope of the recall is adequate, and all that time assumes
that a remedy is available. Thus, while the agency's enforcement
authority is necessary and essential, to identify safety defects, it
does not provide the prophylactic safety effect of standards and
regulations.
Exemption Authority
Advocates is concerned with the NHTSA AV Policy's discussion of the
use of exemption authority and proposed expansion of that authority.
Compliance with safety standards must be maintained to ensure that all
consumers are afforded a minimal level of safety. Advocates concurs
with the agency statement that ``[g]eneral exemptions do not excuse
non-compliance with applicable standards simply because doing so would
be inconvenient or inconsistent with the manufacturers' preferred
vehicle design.'' \53\ However the agency in its discussion regarding
expanding its general exemption authority considers not only increasing
the number of vehicles and duration of the exemptions, but also
mentions the possibility of moving towards relaxed limitations or even
eliminating numerical limits for exempt vehicles altogether.\54\
Advocates strongly opposes this idea which will compromise safety.
---------------------------------------------------------------------------
\53\ NHTSA AV Policy, p. 55.
\54\ NHTSA AV Policy, p. 76.
---------------------------------------------------------------------------
As Advocates' comments on countless past general exemptions have
shown, particularly from the FMVSS, the standard for proving that an
alternative approach represents an equivalent level of safety to the
existing regulatory requirement is murky.\55\ Furthermore, Advocates is
concerned that expansion of the exemption authority would lead to an
alternative path around FMVSS compliance. Advocates concurs with the
agency's expressed concern that ``it would be important to guard
against overuse of the exemption authority which might displace
rulemaking as the de facto primary method of regulating motor vehicles
and equipment.'' \56\ For this reason alone, the expansion of exemption
authority should be avoided. This is not the intent of the agency's
organic statute, its current exemption authority, or the preferred
means by which the agency should ensure public safety. Advocates
recommends that the agency focus on efficiently addressing current
regulatory shortcomings and adapting current requirements through
rulemaking rather than eviscerating the rules through exemptions.
---------------------------------------------------------------------------
\55\ Most exemption requests are supported by little or no
substantive data that meets the equivalent level of safety legal
standard.
\56\ NHTSA AV Policy, p. 76 (emphasis added).
---------------------------------------------------------------------------
IV. Modern Regulatory Tools
Safety Assurance
The discussion of Safety Assurance in the NHTSA AV Policy document
confirms that the NHTSA could establish, by rule, a mandatory Safety
Assurance requirement ``to demonstrate that motor vehicle
manufacturers' and other entities' design, manufacturing, and testing
processes apply NHTSA performance guidance, industry best practices,
and other performance criteria and standards to assure the safe
operation of motor vehicles, before those vehicles are deployed on
public roads.'' \57\ This could apply to level 0-2 vehicles as well as
level 3-5 AVs as well. The agency makes clear that ``NHTSA could
implement many safety assurance tools without additional statutory
authority.'' \58\ Advocates completely agrees and urges the agency to
pursue this course of requiring manufacturers to establish a functional
safety based assessment process to be reviewed by the agency. The
adoption of a required safety assurance process for AVs would still be
consistent with the manufacturer self-certification procedure required
by existing law.\59\ Manufacturers would have to comply with the safety
assessment process and allow NHTSA an opportunity to review the AV
documentation and request additional information. However, following
the agency review the manufacturer would be able to self-certify the
AVs as under current law. The agency would not itself pre-test
prototype AVs to ensure that they conform to the FMVSS and the
verifications contained in the manufacturer safety assessment.
---------------------------------------------------------------------------
\57\ NHTSA AV Policy, pp. 70-71.
\58\ NHTSA AV Policy, p. 71.
\59\ Title 49, U.S.C., Sec. 30115.
---------------------------------------------------------------------------
Pre-Market Approval Authority
Pre-market approval would allow the agency to conduct testing on
prototype vehicles and to ensure no AVs are marketed without approval.
This may be particularly important for highly complex electronic
aspects of vehicle control and to allow examination of computer
programming and logic. However, it is an entirely different regulatory
system that exists currently at the NHTSA. Regardless, the safety
assessment aspect of the agency guidance could be put into place by
rule, without the need for pre-market approval legislation.\60\
---------------------------------------------------------------------------
\60\ The NHTSA's discussion of a hybrid approval process that would
combine self-certification of existing FMVSS and pre-market approval by
the NHTSA, or a third party expert supervised by the agency, for AV
features that are not subject to an FMVSS or regulation, still require
Congress to pass pre-market legislation.
---------------------------------------------------------------------------
Cease and Desist Authority
Advocates strongly supports the enactment of legislation to afford
the NHTSA cease and desist authority to immediately remediate imminent
hazards. This authority, possessed by other regulatory administrations,
including the Federal Motor Carrier Safety Administration, would allow
the agency to expedite taking action in the event that a serious
problem or defect poses an immediate danger to public safety. In a
number of previous and recent safety recalls, the agency might have
invoked such authority to ground vehicles that pose a significant
unreasonable risk to public safety. Imminent hazard authority would
still require some measures of due process and court review to ensure
that the agency does not act impetuously, but would allow a faster
response to address a serious safety problem.
Post Sale Authority to Regulate Software Changes
Advocates agrees with the NHTSA that the agency currently has
authority to regulate software changes that update vehicle programming
and could affect the basis for the original vehicle certification. As
the software governing an AV is part of the vehicle or vehicle
equipment, the agency has authority to regulate changes that could
affect vehicle safety systems and that could result in a defect or give
rise to an unreasonable risk to safety. The agency should require by
regulation any post sale software update be submitted to NHTSA and made
available for public review with a summary of the changes that were
implemented.
Jacqueline S. Gillan,
President.
Joan Claybrook,
Consumer Co-Chair,
Former NHTSA Administrator.
______
Statement of American Car Rental Association
The American Car Rental Association (ACRA) respectfully submits
this statement for the record of the Senate Commerce, Science and
Transportation Committee's hearing on ``Paving the Way for Self-Driving
Vehicles'' on Wednesday, June 14, 2017.
ACRA is the national representative for over 98 percent of our
Nation's car rental industry. ACRA's membership is comprised of over
300 car rental companies, including all of the brands you would
recognize such as Alamo, Avis, Budget, Dollar, Enterprise, Hertz,
National and Thrifty. ACRA members also include many system licensees
and franchisees mid-size, regional and independent car rental companies
as well as smaller, ``mom & pop'' operators. ACRA members have over two
million registered vehicles in service, with fleets ranging in size
from one million cars to ten cars.
ACRA's members strongly support the development and gradual
deployment of ``Highly Automated Vehicles'' (HAVs) to improve
transportation safety and reduce property damage and personal injury
and deaths associated with vehicle accidents. However, the introduction
of HAVs is a complex technical and public policy challenge. This
challenge will require policymakers to address and incorporate existing
safety, consumer protection, privacy and liability issues into a
changing vehicle population that includes HAVs--while at the same time
maintaining flexibility to accommodate new and evolving legal issues
unique to HAVs that may not be apparent today.
The members of the ACRA purchase one out of every nine new cars
sold in the United States each year--almost 2 million vehicles in 2016.
To the extent that HAVs are introduced into the private passenger motor
vehicle fleet in the next decade, ACRA members will be at the forefront
of HAV deployment and on the front lines of the education of car rental
customers with respect to interacting with HAVs safely.
The Promise and Challenges of Autonomous Vehicles--The widespread
introduction of HAVs promises to reduce the number of deaths (about
40,000/year in the United States) and injuries (hundreds of thousands
every year in the United States) caused by motor vehicle accidents--
over 90 percent of which are caused by human error. But this promise is
not without challenges in many complex areas, including thorny public
policy issues that have been debated by many interested parties for
decades, including:
Liability--Federal and State liability statutes generally
hold the driver of a motor vehicle liable for injuries and
property damage caused by that driver's negligence. With
respect to HAVs, there is no ``driver'' per se and thus
determining responsibility for injuries and other harm become
problematic. Federal and State policymakers should consider
assigning liability for accidents caused by HAVs to the
entities most capable of addressing design and functionality
shortcomings in HAVS--in most cases, the vehicle and software
designers and manufacturers, rather than the humans occupying
the vehicle or the fleet owners.
Ownership of Motor Vehicles--As we move towards an era of
widespread HAV deployment, our notions of motor vehicle
ownership likely will undergo a revolutionary change. Instead
of owning our personal automobile, or renting a minivan for a
family vacation, or boarding a bus for a ride to school, or
hailing a taxi--all of these activities may be undertaken with
different types of HAVs which may or may not be owned by an
individual, a school district, or a fleet operator. Resolving
vehicle ownership issues, including maintenance, accident
reporting, data recording and sharing, and other heretofore
unaddressed issues with respect to HAVs will need to be
discussed and resolved.
Taxes and Fees--HAVs hold the promise of eliminating
distinctions between rental cars, taxis, ride-hailing services
and individual motor vehicle ownership. With the introduction
of HAVs, responsibilities must be apportioned for paying
Federal and State motor fuel excise taxes, State and Local fees
on car rentals, ride-hailing services and taxis, and State and
Local vehicle registration and sales taxes.
Harmonization--The customers of ACRA members cross state
lines in their current rental cars without restrictions and
likely will anticipate the ability to do the same with respect
to HAVs rented from ACRA members. As a result, a myriad of
complex and perhaps contradictory State laws or regulations
with respect to technical, safety or operational standards for
HAVs should be avoided wherever possible. Continued State
regulation of HAVs in traditional areas such as licensing,
registration and insurance requirements would not in most
instances pose impediments to the introduction of HAVs in
ACRA's opinion.
Privacy--Federal and State regulators have started to
wrestle with the difficult challenges of maintaining individual
privacy with respect to data generated by today's increasingly
complex and technologically advanced motor vehicles and
promoting transportation safety and enforcement of Federal and
State laws. Such thorny privacy issues will only be multiplied
with HAVs, and ACRA urges policymakers to preserve the right of
vehicle owners to control and own the data generated by HAVs.
Cybersecurity--The increased automation of motor vehicles,
leading ultimately to deployment of HAVs, heightens the risk of
cyber-attacks on single cars or groups of vehicles. Such risks
must be managed by vehicle manufacturers and designers.
However, the same technology that opens HAVs to cyber-attacks
may hold the promise of reducing motor vehicle theft and other
crimes involving vehicles. The cybersecurity issues related to
HAVs must be balanced between protection of the vehicle's
occupants and aiding law enforcement agencies in crime
prevention and the apprehension of criminals.
``Training'' of HAV Users--Unless an individual purchases an
HAV and is briefed by the seller on the operation of the
vehicle (much as is done at many vehicle dealerships today when
a vehicle is purchased), most consumers' first interaction with
an HAV likely will be in a car rental transaction or other
short-term use scenario such as a taxi or ride-hailing use.
Manufacturers of the HAVs' hardware and software must assume
responsibility for providing consumers with a training manual
or user interface presentation upon entry into the HAV. Placing
this training requirement on the manufacturer will be much more
efficient and effective than a requirement that the owners--
whether individuals or fleet management companies--of the HAVs
provide this training on the use and operation of the HAV.
Special Attention to Data/Information Ownership and Control--ACRA
posits that the owners of HAVs--whether an individual, a corporate
entity, or a fleet management company--and their agents (such as
property and casualty insurers) must retain full, real-time access and
control over the personal and vehicle information associated with and
generated by HAVs.
Such information access and control must include data ``read'' and
``write'' capabilities, as well as authenticated, remote command and
control of a stationary vehicle. In addition, owners of HAVs with
multiple users, such as consumers with multiple drivers in the same
family, or car rental companies with many customers using one HAV, must
be provided with a secure and simple means of erasing personal
information from the info-entertainment systems of all vehicles,
including HAVs.
Currently, the owner of a motor vehicle has full ownership over the
personal information stored in that vehicle and the vehicle information
generated by and stored by the vehicle. Such a data ownership construct
must continue as we transition to HAVs to protect consumer privacy, to
ensure that the safety promise of HAVs is fully realized, to promote
competition, and to permit owners of HAVs to manage the information
generated by their vehicles--whether it is one HAV owned by an
individual or 100,000 HAVs owned by a car rental company.
Some stakeholders may suggest that access to personal and vehicle
data must be limited to hardware or software providers in order to
fully deploy HAVs. Such a closed silo of access to vehicle information
would represent a complete public policy reversal of decades of
consumer and privacy protection. The proponents of such a closed data
system have not advanced convincing arguments as to why they should be
granted a monopoly on this data--a monopoly that would be in place even
when the hardware or software company doesn't own the HAV in question.
ACRA looks forward to working with the Committee and all
stakeholders to formulate a personal and vehicle information and data
access public policy for HAVs that is consistent with consumer
protection and privacy principles while at the same time achieving the
safety goals promised by HAVs.
Thank you for the opportunity to present this statement for the
record at this hearing. ACRA stands ready to work with the members of
the Committee and all State and Federal legislators and regulators, as
well as the many stakeholders interested in the development and
introduction of HAVs, in the months and years ahead.
______
Statement from Continental
Continental is a leading Tier 1 supplier that develops intelligent
technologies for transporting people and their goods. We provide our
automotive customers with sustainable, safe and affordable solutions
that enhance automotive safety. In 2016 we generated more than $43
billion in sales within our five divisions, Chassis & Safety, Interior,
Powertrain, Tires, and ContiTech. Continental employs more than 20,000
employees in the U.S. at more than 80 facilities located in 27 states
and has more than 220,000 employees in 55 countries worldwide.
In 2015 there were more than 35,000 lives lost in the U.S. due to
traffic crashes. Projections for 2016 are expected to increase to more
than 40,000 fatalities, a level we haven't seen in a decade. While this
is an alarming number, it is even more startling at a global level--
more than 1.2 million people die in roadway crashes and another 50
million are injured. This is unacceptable and reversing this trend is
what motivates each and every employee at Continental.
In the last 45 years the U.S. has experienced a relative declining
trend in traffic fatalities with respect to an increased number of
vehicles on the road and number of miles driven. This is due in large
part to improved vehicle safety technologies. In the early 1970s the
number of injuries and fatalities were at an all-time high. The
introduction of the seat belt helped to reduce the total number of
traffic fatalities by 10,000 in a few short years. In 1983, the number
of fatalities was the lowest in 20 years due to the introduction of
anti-lock braking systems. As numbers began to rise again, the airbag
became standard in vehicles reducing injuries and fatalities down to
its lowest number in 30 years. The introduction of electronic stability
control in the mid-1990s helped to reduce traffic accidents to the
lowest number in 50 years. Continental projects new crash-avoidance
technologies will once again reverse the recent increase in fatalities
as the auto industry moves toward a more widespread implementation of
Advanced Driver Assist Systems (ADAS).
Innovation has always been at the heart of the automotive industry.
From the original concept of the automobile in the late 1800s, the mass
production lines pioneered in Detroit, to today, the automotive
industry has always invested in research and development to make their
products safer, more reliable and more affordable. Today, we are
witnessing the automotive industry evolve from a crashworthiness
mindset, where manufacturers try to make the passenger cabin more
survivable in the event of an accident towards a crash avoidance
mindset--after all, the best way to survive a crash is to avoid one in
the first place.
Continental, and our dedicated employees, are committed to
developing Safe and Dynamic Driving technologies towards Vision Zero.
Vision Zero means a future with zero traffic fatalities, injuries and
ultimately zero accidents. Such a future can only be achieved with the
help of innovative active and passive safety, driver assistance, and
automated driving technologies. As Continental brings these
technologies to market, we exhaustively test products, and subsystems,
as part of a larger system of advanced driving assistance technologies
that will be integrated with a variety of components by original
equipment manufacturers.
Our Vision Zero philosophy is embedded in each technology we
develop as we continue to enable automated driving. At Continental, we
describe our systems approach through three primary actions--sense,
plan, and act. Whether the technology simply assists the driver like
many systems on the road today, or ultimately takes over the driving
task completely, it first must SENSE the surrounding environment and
gather the necessary data that can be interpreted. Sophisticated sensor
systems can help eliminate human error and distractions by providing
360-degree awareness of the road at all times. The data gathered from
the sensors is then analyzed to identify obstacles or hazards. Our
systems then dynamically develop a PLAN to determine how to assist the
driver. Once that plan is in place, the systems will ACT to execute the
plan to safely and comfortably pilot the vehicle and in certain cases
avoid a hazard or crash situation. Our Sense, Plan, Act approach is the
foundation behind Continental's active safety and Advanced Driver
Assistance System technology, and is a key component to advancing
automated driving systems. We believe that when fully automated driving
is possible, traffic fatalities can be reduced by 90 percent because
that is the percentage of accidents that are caused by human error.
Continental has been an active participant globally in policy
discussions and initiatives with governments, automotive industry
partners, trade associations and other standard setting organizations.
The collaborative efforts to help establish consistency within the
emerging self-driving market has been crucial to the advancement of
automated driving technologies. Continental is currently engaged with
the Department of Transportation's Smart Cities Program. Several of our
divisions are working together to develop a highly sophisticated
intersection in Columbus, Ohio, with vehicle and integrated
infrastructure technologies that will help save the lives of vehicle
occupants as well as pedestrians while improving transportation
efficiency in urban environments. We support the National Highway
Traffic Safety Administration's recent adoption of the SAE
International definitions of automation, as we believe it is beneficial
to helping educate the public in order to distinguish between different
automated technologies and garner public acceptance.
Continental is one of the leading suppliers in this market, with a
complete portfolio of technologies for all defined levels of
automation. Each innovative safety feature undergoes an extensive
testing process before becoming available to the market. As a supplier,
we currently develop a multitude of innovative technologies that can
save lives and enhance the driving experience under the Level 0 to
Level 2 definitions of automation. These products are designed based on
the needs of our customers to assist the driver in interpreting the
surrounding environment and control the vehicle in order to prevent an
accident from occurring.
Continental has been integral in the deployment of current crash
avoidance technologies such as lane keep assist, rear back up assist,
automatic emergency braking, and adaptive cruise control, to name a
few. These crash avoidance technologies are the building blocks to
higher levels of automated driving and need to be embraced as crash
avoidance technologies that save lives. All of these technologies can
be found throughout the fleets of most vehicle manufacturers.
As the industry moves forward towards Level 3 automation technology
and beyond, Continental is positioned to supply public and personal
transportation needs with the safest and most advanced technology
available on the market. The world and the behavior of drivers within
it are ever changing, and the vehicle must adapt to these changing
trends. Our children seem to rely on smartphones more so than vehicles.
Living in a world of distractions has become commonplace. Automotive
technology must be developed accordingly. That is why Continental has
put a great deal of effort into Human Machine Interface technology. We
want the driver to be aware of their surroundings, be aware of what the
systems in the vehicle are doing, and be aware of when it is safe to
relinquish control of the vehicle and when to reengage with the
vehicle. In addition to informing the occupants, keeping them safe, and
pedestrians safe, we must also secure the systems within the vehicle.
As part of system development for Highly Automated Driving, we focus on
redundancy of vehicle safety systems. That is why we are developing
complimentary systems and technologies that support existing safety
systems in the vehicle's architecture.
Since 2011, we have continued a pursuit of testing and developing
highly automated driving with next generation technologies like
automated parking, cruising chauffeur and a complete self-driving
vehicle in combination with V2V/V2X technology. We were the first
supplier in the U.S. to be awarded a testing license for automated
vehicles in Nevada and are currently testing our third generation
automated vehicle on highways and roads throughout the country and
around the world. We are currently integrating sophisticated
technologies such as high resolution flash lidar, which will expand the
vehicle's detection capabilities. This is the same technology that has
been deployed on space shuttles at the most advanced technical level,
and we are working to utilize its potential for road applications. But,
our continued efforts in this direction would benefit greatly from an
investment in infrastructure that promotes vehicle to X communication,
a dedicated spectrum communication band that can be utilized by current
and future safety systems, and harmonization of safety laws that allows
for the full real world testing of these technologies.
The challenges in broadly testing this new and innovative safety
technology across the country are great. The industry currently faces
considerable uncertainty on state and Federal requirements that would
require clarification from the Federal Government's exclusive authority
to regulate all motor vehicles. The safe commercial deployment of
potential life saving technology depends on the ability to extensively
test on public roads under all conditions. In order to envision a
future of full automation, the government must review Federal motor
vehicle safety standards that would allow for vehicles that may not be
under the full control of a driver at all times. Similar to the need of
improved road conditions as automobiles transitioned from rural
landscapes to metropolitan areas in the early 1900s, we need a road
infrastructure that complements automotive advancements, and a legal
framework that supports a new system of mobility.
The automotive world is one of excitement. Software developers are
becoming automotive suppliers, automotive companies are becoming
software developers, and our vehicles are becoming our smart-device.
The world of mobility has the capability of expanding to unimaginable
independence and personal freedom without sacrificing the safety of
future generations. Continental stands at the ready, alongside our
industry colleagues, to work with the Committee and Congress in helping
construct laws that foster innovation, enable mobility, and create a
safer environment for the public.
______
Statement of Property Casualty Insurers Association of America
The promise of ``self-driving'' vehicles to improve road safety and
mobility continues to generate debate about the appropriate regulatory
framework for the testing and deployment of such vehicles. While
determining how our existing vehicle regulatory systems need to change
is very important, there is a fundamental mismatch between the public
perceptions that auto accidents and insurance costs are decreasing with
the stark reality that our roads are becoming increasingly dangerous
resulting in rising costs. Auto accident deaths have increased by 14
percent over the last two years, and the number of people injured on
our roads has increased by more than 12 percent since 2014. Someday in
the future self-driving cars may reduce the number of accidents and
deaths. However, the potential of automated vehicle technology stands
in sharp contrast to what is happening on our roads today. The Property
Casualty Insurers Association of America (PCI) is composed of 1,000
member companies, representing the broadest cross section of insurers
of any national trade association. PCI members write $202 billion in
annual premium, 35 percent of the Nation's property casualty insurance.
That figure includes over $97 billion, or 42 percent of the auto
insurance premium written in the United States. PCI's has analyzed the
recent increase in auto insurance claim frequency and found strong
correlations with traffic congestion and distracted driving, weaker
correlations from increasing populations of novice and older drivers,
and some correlation with liberalized marijuana laws. While it is
important to prepare for the automated vehicle of the future, we urge
policymakers to continue to focus on the auto safety challenges that
face us today such as distracted and impaired driving. The FAST Act
provides for continuing efforts to increase public awareness and
improving enforcement as well as establishing an enforceable impairment
standard for drivers under the influence of marijuana. These are
critically important measures for reducing accidents, injuries and
deaths on our Nation's roads. The importance of addressing these issues
was also the subject of a 2016 bipartisan letter from 23 members of
congress urging prompt implementation of these provisions of the FAST
Act by the Department of Transportation.
Increasing automation of driving functions will mean that some
motor vehicle laws and regulations will need to be changed to
accommodate the testing and deployment of self-driving cars. Testing
requirements, guidelines and standards for use on public roads should
set clear expectations for the public and provide clear direction for
technology developers and manufacturers for compliance. Modifications
to existing auto safety laws and motor vehicle safety standards should
be rare and limited to only the highest levels (i.e., fully autonomous)
of automated driving and should clearly define the levels of automation
to which the modification applies. Clear and effectively enforced auto
safety laws and vehicle standards can save lives on our roads today
and, when applied to automated driving systems, develop public
confidence that will ultimately determine if the technology realizes
its potential.
The National Highway Traffic Safety Administration's (NHTSA)
``Federal Automated Vehicle Policy,'' is intended to provide guidance
for states on the testing and deployment of highly automated vehicles
(HAV's). This policy guidance briefly references insurance yet does
raise issues that are important to the automobile insurance market as
it seeks to adapt and develop new products to meet consumer's needs
that should be considered in any policy discussion of self-driving
vehicles.
Recognition of State Regulation of Insurance and Liability Issues
NHTSA's policy identifies as Federal responsibilities: setting and
enforcing safety standards for motor vehicles, recalls, promoting
public awareness and providing guidance for the states. NHTSA's policy
also recognizes that it is the state's role to license drivers and
vehicles, enforce traffic laws and regulate motor vehicle insurance,
tort and criminal liability issues as they pertain to automated
vehicles. PCI shares the view that the states should continue to have
primacy on motor vehicle insurance and liability issues as they do
today, and we support NHTSA's recognition of that role.
NHTSA's policy also repeats the recommendation from its 2013
guidance that entities testing automated technology should provide
proof of financial responsibility coverage of at least $5 million. PCI
has not taken a position on this coverage requirement. But as highly
automated vehicles (HAV's) are deployed for public, states will need to
consider what, if any, changes need to be made to the states existing
motor vehicle financial responsibility laws.
Data Collection and Access
As policymakers consider what data should be collected and retained
by automated vehicles it is essential for providing customer service
that the rules provide for reasonable access to that data for insurers
for claims handling and underwriting purposes. In many auto accidents,
apportionment of liability is likely to hinge upon whether a human
driver or the vehicle itself was in control and what actions either the
driver or the vehicle did or did not take immediately prior to the loss
event. Access to data for insurers will speed claims handling and
potentially avoid disputes that could delay compensation to accident
victims. Access to historical anonymized data on the different
automated vehicle systems will also be important to help insurers
innovate and develop new insurance products as the nature of the risk
changes.
Conclusion
Automated driving technology holds great promise for the future,
and implementing clear policies on the Federal and state roles in
regulating automated vehicle technology and ensuring that insurers have
access to vehicle data on reasonable terms to efficiently handle
claims, develop products and underwriting methods are an essential
first step toward that future. However, policymakers must not lose site
of the auto safety issues that face us today. We look forward to
working with policymakers at the Federal and state level to reduce
accidents on our roads today and in future.
______
Uber
June 13, 2017
Hon. John Thune, Chairman,
Committee on Commerce, Science, and Transportation,
United States Senate,
Washington, DC.
Hon. Bill Nelson, Ranking Member,
Committee on Commerce, Science, and Transportation,
United States Senate,
Washington, DC.
Dear Chairman Thune, Ranking Member Nelson, and members of the
Committee:
We appreciate the opportunity to provide written testimony on the
future of self-driving or automated vehicle technology.
Self-driving cars have the potential to bring unimaginable benefits
to cities, and drastically improve quality of life for millions of
people around the world. But it is a relatively young and rapidly
changing field. How policymakers choose to regulate this space will
play a critical role in determining whether the full potential of this
technology will be realized.
About Uber
Uber is a technology company with a simple vision: make safe and
affordable transportation available everywhere, for everyone and
everything at the push of a button. Since our founding, we have worked
to change the way people think about mobility, and continue to use
technology to connect people and goods to places. Today, 75 percent of
the U.S. population lives in a county where they have access to Uber.
Two years ago, we established our Advanced Technologies Group in
Pittsburgh, Pennsylvania. And today, we employ one of the strongest
self-driving engineering groups in the world, with the practical
experience that comes from 1.2 billion miles traveled on roads every
month through our ridesharing and delivery services in hundreds of
cities, and world-class manufacturing partners.
Our goal is to make safe self-driving cars available to people
everywhere. After thorough testing and evaluation in closed settings--
and on-road testing with employees--we began operating Self-Driving
Ubers on public roads in Pittsburgh last September, giving uberX riders
an opportunity to experience self-driving technology for themselves. In
February, our Self-Driving Ubers started picking up passengers in
Tempe, Arizona.
Before any rider experiences our technology, we validate it in
simulation, monitor it on a closed test track, and test it on-road
without passengers. In addition, we ensure that trained operators are
in the driver's seat for each release of the nascent technology to
continually monitor and learn from the performance of the system. We
believe that the deployment of self-driving cars will only succeed if
we are able to learn from real-word situations. It also provides us
with an opportunity to gain and preserving user trust. Through our
Pittsburgh and Tempe operations, we are accomplishing both.
While we have not yet deployed fully self-driving cars--i.e.
vehicles that operate without a driver--we hope to do so in the coming
years. Our hope is that by transitioning to fleets of self-driving
cars, we can help save more than a million lives each year around the
world, reduce parking and traffic on roads, open up more land for
commercial and residential development, reduce carbon emissions
overall, and reduce the cost of transportation for average
households.\1\,\2\ According to the University of Texas,
these self-driving-led improvements could lead to cost savings of over
$1.4 trillion per year in avoided car crashes, reduced congestion
delays, and productivity gains en route.\3\
---------------------------------------------------------------------------
\1\ ITF study, SDV future requires vehicle fleet less than 10
percent of its current size: http://www.itf-oecd.org/sites/default/
files/docs/15cpb_self-drivingcars.pdf
\2\ UC Berkeley study, SDVs with EVs could reduce emissions per
vehicle mile by more than 90 percent https://drive.google.com/a/
uber.com/file/d/0BzdDAOtdz76gUTFlX0x1bERnN2M/view?usp=sharing
\3\ https://mobilitylab.org/2017/03/20/autonomous-vehicles-safety-
add-traffic/
---------------------------------------------------------------------------
To unlock these outcomes, it's critical that our regulatory
environment does the following:
1. Enables and encourages shared vehicle operation and use (through
ridesharing or commercial vehicle fleets) in addition to the
traditional direct-to-consumer sales model for passenger
vehicles;
2. Acknowledges the differences between self-driving software and
traditional vehicle equipment software and sets the appropriate
rules for each;
3. Does not fundamentally re-balance powers between local and
Federal authorities that regulate vehicle safety or tie the
industry to standards that are still evolving.
Encouraging shared vehicle operation and use
To date, policies addressing passenger vehicle operation and use
have largely been based on a traditional manufacturer-sold, owner-
operated model. But that is not the approach that many technology
companies and automakers are taking for self-driving cars. Rather,
companies are investing in shared, self-driving fleets because the
model creates far more potential for the faster and safer development
of self-driving technologies for several reasons:
First, the sheer expense of owning and operating a self-
driving car, which is expensive and difficult to build, makes
the technology cost-prohibitive for most households. The vast
majority of people will encounter, experience, and become
familiar with this technology through shared self-driving
fleets that are deployed through a ridesharing network. In
addition to gaining access at a more affordable cost, a fleet
approach will also provide access to higher quality sensing and
computing features that will be continually updated.
Second, a world where people simply trade the cars they have
today for self-driving ones will neither reap the benefits of
less congestion and pollution, nor eliminate the expense of
owning a car. The International Transport Forum has studied the
potential of shared self-driving cars, and proposed that a city
that moves to a shared, self-driving future will require a
vehicle fleet less than 10 percent its current size.\4\
---------------------------------------------------------------------------
\4\ http://www.itf-oecd.org/sites/default/files/docs/15cpb_self-
drivingcars.pdf
Third, fleet operators will be better equipped to ensure the
safe use of self-driving cars than individual owner-operators.
In a direct-to-consumer model, an owner-operator would be
responsible for maintaining hardware and downloading software
updates to improve the safety of his or her vehicle. He or she
would also be responsible for making decisions about when,
where, and how to operate the vehicle. By contrast, a fleet
operator has the ability to hire qualified professionals,
perform vehicle maintenance on a daily basis if necessary,
update both software and hardware, and impose operational
restrictions on use of all the vehicles in the fleet. For
example, if a fleet operator understood that a particular route
was difficult for self-driving cars, that operator could
mandate navigational paths around that route; similarly, an
operator could restrict the use of self-driving cars during
---------------------------------------------------------------------------
difficult weather conditions.
Finally, a fleet approach allows companies to control how
cars are deployed to ensure they only operate in conditions
that the cars are capable of handling. Companies are then able
to adjust the operating conditions and improve the software as-
needed. Today, Self-Driving Ubers only run along certain roads
that have been pre-selected based on software capabilities and
corresponding road conditions; these roads have been mapped in
advance and programmed to take road signs and other
infrastructure into account. By contrast, self-driving cars
sold to individuals must be able to operate on all roads and
under all conditions before they are distributed.
Our decision to operate self-driving cars using a ridesharing
fleet, rather than following the end-consumer model traditionally used
by automotive manufacturers, is a natural byproduct of these facts.
To be clear, this vision of the future does not reflect a dispute
between technology companies and traditional automakers. Many
traditional automakers have also indicated they are interested in
pursuing various types of fleet models--such as Mercedes Boost and GM
Maven. So as policymakers consider legislation addressing self-driving
and automated technologies, it is critical they take the benefits of a
paradigm shift from private car ownership to shared self-driving cars
into account.
Self-driving software cannot be regulated the way other vehicle
equipment is
The embedded software (e.g., Ford SYNC or cruise control
technology) in today's motor cars sold to end consumers is installed
once and rarely--if ever--updated. By contrast, self-driving software
(e.g., maps, routes, sensing and perception) must be updated frequently
for safety and security. NHTSA's guidance last year demonstrated a
desire to regulate self-driving technology under its authority over
motor vehicle equipment and embedded software, but the proposed
approach overlooks the fundamental difference between self-driving
software and the software traditionally embedded in motor vehicle
equipment.
A regulatory development and enforcement process that takes months
or years to complete may be sensible for embedded software that cannot
be easily updated after a vehicle is sold. But this process should not
prohibit companies from upgrading their self-driving software daily.
Given the speed of our learnings, and the safety/security concerns each
software update may address, voluntary standards imposed by industry
groups will prove more effective than regulatory enforcement (more on
those standards below). Overly burdensome restrictions on software
updates will only reduce the pace at which safety/security improvements
are deployed, and NHTSA is neither equipped nor staffed to keep pace
with how rapidly these updates must roll out. Ultimately, this will
prevent the safest technology from making its way onto the roads as
quickly as possible.
As a parallel example, most technology companies already have
security teams focused on identifying vulnerabilities and preventing
them from being exploited by patching each flaw quickly after it is
discovered. If a software review process is required each time a
vulnerability in self-driving software is discovered, that would impede
developers' abilities to keep self-driving cars secure by pushing
timely updates.
While we appreciate that NHTSA's proposed process (Federal
Automated Vehicles Policy (Docket No. NHTSA-2016-0090) may have been
motivated by a goal to develop a flexible framework that would adapt to
the rapidly changing technology, we believe it will have the opposite
effect. Their approach treats this software as one would a rigid
product with infrequent (if any) changes, instead of a constantly
evolving technology. It also does not take into consideration the idea
that self-driving cars may only be deployed in certain pre-defined
scenarios that companies deem their software ready to address (e.g.,
specific routes). Instead of establishing a burdensome oversight
process, the agency's focus should be on defining what results
companies must achieve for a specific use-case instead of how they are
achieved.
Self-driving technology does not warrant re-balancing state and Federal
powers
In the United States, Federal authorities have always regulated
motor vehicle equipment, while regulating drivers who operate vehicles
has always been in the purview of individual states. The states, as a
result, have long been an incubator for innovative regulatory
approaches to improve driver safety. For example, states have enacted
laws combating driver distraction by setting rules around cell phone
use and texting, which apply to the driver's behavior while operating
the vehicle. States determine what rules of the road are most
appropriate for their constituents and their environment, and tailor
their own regulation of driver behavior and vehicle use to meet local
needs.
Uber has experienced first-hand how innovation at the state and
local level allows for the creation of rules tailored to the local
environment. We were able to work with regulators in Florida to develop
a framework that suited the state's needs--one that was ultimately very
different from the framework used in New York City, and different still
from the one used in Chicago. As a result, within five years over 75
percent of the U.S. population has access to ridesharing, and has
enjoyed the benefits that come with greater access to affordable,
reliable transportation.
Similarly, as self-driving software begins to take the role of the
human driver, states should be allowed to continue to investigate,
innovate, and pilot different approaches as they deem appropriate.
States have always defined the standard by which humans are licensed to
drive on state roads, and evaluate a human's ability to maneuver
driving tasks. For example, some (but not all) states require drivers
to demonstrate an ability to parallel park a vehicle. In keeping with
that approach, states should evaluate a self-driving car's ability to
successfully maneuver those tasks.
However, some entities have argued that self-driving cars will
cross state borders and that a national framework is necessary. We
believe that self-driving software should be able to determine a
vehicle's location and adapt to local requirements as needed--
particularly if there are specific local dynamics at play like
inclement weather conditions. In addition, we recognize the role of
Congress and NHTSA in setting high-level, universal standards, and look
forward to working with officials and other industry stakeholders on
that process.
Relatedly, industry itself often quickly settles on best practice
safety standards in working groups. For example, ISO 26262, an
international standard for road vehicle functional safety, is currently
being updated to address self-driving technology. In addition, a new
standard called SoTIF or Safety of the Intended Function has been
drafted specifically to address the complexities of self-driving
systems. And as former Administrator Rosekind himself highlighted
during the November 16, 2016 U.S. Senate Appropriations Committee
hearing on ``The Automated and Self-Driving Vehicle Revolution,'' over
20 automakers agreed to emergency braking features well ahead of the
regulatory requirement, and industry came together to develop best
practices on cybersecurity. This approach is preferable, not only
because technology changes so rapidly, but also because it ensures one
technology is not favored over another.
This is also why commonly-used standards like the levels of
automation set by the SAE must be allowed to shift as technology
changes--with industry input--rather than be codified in law. The SAE
levels were intended to describe the various levels of automation so
that developers could self-classify the types of technology they were
building. They were never intended--and were not drafted--to be binding
legal standards. Companies are already working simultaneously on
technologies that fall under different levels of automation. For
example, refining street maps and routing information for human drivers
to use while operating vehicles that require human input and monitoring
does not preclude a company from developing high-definition maps and
autonomy software that allows vehicles to perceive their surroundings
and navigate without human driver input. Multiple investments allow
companies to leapfrog certain levels as they become obsolete. It's
possible that in the future, none of the levels of automation will
require human input as technology advances.
Recommendations
Even under relatively optimistic scenarios, we likely will not see
self-driving cars become a significant portion of the U.S. vehicle
fleet for many years. In the coming years, 17 million new vehicles will
be sold per year, only a small number of which will be self-driving
cars; self-driving cars will only be a small proportion of the overall
U.S. fleet of over 250 million vehicles for years to come. These facts
suggest that--despite the heavy interest among regulators, media, and
the public--elaborate guidance or hasty regulation of self-driving cars
would be inappropriate. The tremendous potential of this technology--
for instance, to remove the human choice or error tied to 94 percent of
crashes--can best be achieved through alternative approaches.
Since conventional vehicles will occupy the vast majority of new
vehicles for the foreseeable future, we encourage Congress and the DOT
to focus their attention and limited agency resources on higher-impact
efforts that can serve all road users regardless of levels of
automation. For example, ensuring that funding is going toward adopting
Complete Street Design Standards, building uniform pavement without
potholes, and creating protected bike lanes or separations for
pedestrian walkways that calm traffic and reduce potential interactions
between vehicles and bikers or pedestrians.
To be clear, a dramatic `smart infrastructure' overhaul is not
necessary for self-driving deployment. Rather, Congress should focus on
doing what it already does best: maintaining our road infrastructure,
which has been underfunded for some time, and needs $3.6 trillion in
repairs by 2020.\5\ In fact, focusing too heavily on `smart
infrastructure' could generate costly investments and will slow down
the adoption of self-driving cars and limit their use to only those
places able to afford costly connected infrastructure adoption.
---------------------------------------------------------------------------
\5\ American Society of Civil Engineers, http://
www.infrastructurereportcard.org/
---------------------------------------------------------------------------
Finally, to the extent Congress and NHTSA choose to allocate
resources specifically to self-driving matters now, we recommend that
the agency investigate amending outdated FMVSS to address self-driving
cars. Nearly half of the FMVSS have some reference to the driver,
driving controls, or driver seating position, and NHTSA has not laid
out a time-frame for amending them. If they remain unchanged, these
likely will be an impediment to the development and design of self-
driving cars, make product planning more difficult, and potentially
force companies to include equipment that could make self-driving cars
less safe.
Uber thanks Chairman Thune, Ranking Member Nelson, Senator Peters
and other members of the Committee for highlighting the importance of
policy frameworks that enable companies able to deploy self-driving
cars safely and rapidly. We appreciate the opportunity to provide input
on how governments and companies can work together to maximize the
benefits of this technology, and look forward to working with members
of the Committee and others to make it a reality.
Sincerely,
Eric Meyhofer,
Head of Advanced Technologies Group (ATG),
Uber Technologies, Inc.
cc: Hon. Gary Peters, Commerce Committee Member
______
March 29, 2016
Moody's forecasts U.S. adoption of self-driving cars `several decades'
away
Reuters Staff
WASHINGTON (Reuters)--Near-universal adoption of self-driving cars
in the United States is likely by around 2055 and poses a long-term
threat to auto insurance companies, Moody's Investors Service said in a
research report released Tuesday.
The report forecasts that fully autonomous or driverless cars could
be a common option on U.S. vehicles by about 2030 and is likely to be
standard on all new vehicles sold around 2035.
That would lead to the technology being in a majority of cars on
U.S. roads by around 2045.
``Accident frequency will fall sharply over time, and will
ultimately translate into significantly
lower premiums and consequently lower profits for auto insurers,''
the report said.
The Moody's report estimates that over the next five to 10 years,
automotive technologies like automatic emergency braking are likely to
boost insurance profitability as they may reduce crashes but not lead
to an immediate reduction in rates.
http://www.reuters.com/article/us-moody-s-autos-selfdriving/moodys-
forecasts-u-s-adoption-of-self-driving-cars-several-decades-away-
idUSKCNOWV220
______
Electronic Privacy Information Center
Washington, DC, June 13, 2017
Hon. John Thune, Chairman,
Hon. Bill Nelson, Ranking Member,
U.S. Senate Committee on Commerce, Science, and Transportation,
Washington, DC.
RE: Hearing on ``Paving the Way for Self-Driving Vehicles''
Dear Chairman Thune and Ranking Member Nelson:
We write to your regarding the upcoming hearing ``Paving the Way
for Self-Driving Vehicles,'' \1\ on the privacy and safety risks of
connected and autonomous vehicles. For more than a decade, the
Electronic Privacy Information Center (``EPIC'') has warned Federal
agencies and Congress about the growing risks to privacy resulting from
the increasing collection and use of personal data concerning the
operation of motor vehicles.\2\
---------------------------------------------------------------------------
\1\ Paving the Way for Self-Driving Vehicles, 115th Cong. (2017),
S. Comm. on Commerce, Science, and Transportation, https://
www.commerce.senate.gov/public/index.cfm/press
releases?ID=B7164253-4A43-4B70-8A73-68BFFE9EAD1A (June 14, 2017).
\2\ See generally EPIC, ``Automobile Event Data Recorders (Black
Boxes) and Privacy,'' https://epic.org/privacy/edrs/. See also EPIC,
Comments, Docket No. NHTSA-2002-13546 (Feb. 28, 2003), available at
https://epic.org/privacy/drivers/edr_comments.pdf (``There need to be
clear guidelines for how the data can be accessed and processed by
third parties following the use limitation and openness or transparency
principles.''); EPIC, Comments on Federal Motor Vehicle Safety
Standards; V2V Communications, Docket No. 2016-0126 (Apr. 12, 2017),
https://epic.org/apa/comments/EPIC-NHTSA-V2V-Communications.pdf
[hereafter ``V2V comments'']; EPIC, Comments on the Benefits,
Challenges, and Potential Roles for the Government in Fostering the
Advancement of the Internet of Things, Docket No. 160331306-6306-01
(June 2, 2016), https://epic.org/apa/comments/EPIC-NTIA-on-IOT.pdf;
EPIC, Comments on Federal Motor Vehicle Safety Standards: ``Vehicle-to-
Vehicle (V2V) Communications,'' Docket No. NHTSA-2014-0022 (Oct. 20,
2014), https://epic.org/privacy/edrs/EPIC-NHTSA-V2V-Cmts.pdf; EPIC,
Comments on the Privacy and Security Implications of the Internet of
Things (June 1, 2013), https://epic.org/privacy/ftc/EPIC-FTC-IoT-
Cmts.pdf; EPIC et al., Comments on the Federal Motor Safety Standards;
Event Data Recorders, Docket No. NHTSA-2012-0177 (Feb. 11, 2013),
https://epic.org/privacy/edrs/EPIC-Coal-NHTSA-EDR-Cmts.pdf; EPIC,
Comments, Docket No. NHTSA-2004-18029 (Aug. 13, 2004); https://
epic.org/privacy/drivers/edr_
comm81304.html.
---------------------------------------------------------------------------
EPIC was established in 1994 to focus public attention on emerging
privacy and civil liberties issues. EPIC engages in a wide range of
public policy and litigation activities. EPIC testified before the
House of Representatives in 2015 on ``the Internet of Cars.'' \3\
Recently, EPIC urged the Ninth Circuit of Appeals to protect the right
of consumers to pursue safety issues with connected vehicles.\4\ As
EPIC explained in that case:
---------------------------------------------------------------------------
\3\ Statement of Khaliah Barnes, hearing on The Internet of Cars
before the House Committee on Oversight and Government Reform, November
18, 2015, https://epic.org/privacy/edrs/EPIC-Connected-Cars-Testimony-
Nov-18-2015.pdf.
\4\ Brief of Amicus Curiae EPIC, Cahen v. Toyota Motor Corporation,
No. 16-15496 (9th Cir. Aug. 5, 2016), https://epic.org/amicus/cahen/
EPIC-Amicus-Cahen-Toyota.pdf.
``Connected vehicles'' expose American drivers to the risks of
data breach, auto theft, and physical injury. The internal
computer systems for these vehicles are subject to hacking,
unbounded data collection, and broad-scale cyber attack.
Despite this extraordinary risk, car manufacturers are
expanding the reach of networked vehicles that enable third
party access to driver data and vehicle operational systems.\5\
---------------------------------------------------------------------------
\5\ Id.
EPIC will also be participating in the FTC/NHTSA workshop on
privacy and security issues in connected cars later this month.\6\
---------------------------------------------------------------------------
\6\ Connected Cars: Privacy, Security Issues Related to Connected
Automated Vehicles, Federal Trade Commission, https://www.ftc.gov/news-
events/events-calendar/2017/06/connected-cars
-privacy-security-issues-related-connected.
---------------------------------------------------------------------------
Connected vehicles pose substantial safety and privacy risks. To
date there have been several high-profile accidents involving self-
driving car including:
A bicyclist was struck by an autonomous driving car after it
suddenly activated its brakes;\7\
---------------------------------------------------------------------------
\7\ Patrick May, Robot-Human Smackdown: Self-Driving Car and
Bicyclist Collide in San Francisco, The Mercury News, Jun. 9, 2017,
http://www.mercurynews.com/2017/06/09/robot-
human-smackdown-self-driving-car-and-bicyclist-collide-in-san-
francisco/.
Uber recently suspending its ``self-driving'' program in
Arizona after one of the company's vehicles struck a car with
passengers inside.\8\ The Uber vehicle was in self-driving
mode, presumably ``Level 3'';
---------------------------------------------------------------------------
\8\ Mike Isaac, Uber Suspends Tests of Self-Driving Vehicles After
Arizona Crash, New York Times, Mar. 25, 2017, https://www.nytimes.com/
2017/03/25/business/uber-suspends-tests
-of-self-driving-vehicles-after-arizona-crash.html; Steven Overly, Uber
Self-Driving Car Flipped On Side In Arizona Crash, Chicago Tribune,
Mar. 25, 2017, http://www.chicagotribune
.com/bluesky/technology/ct-uber-self-driving-car-crash-20170325-
story.html.
A self-driving car failed to stop at a red light at a busy
intersection;\9\ and
---------------------------------------------------------------------------
\9\ Mike Isaac & Daisuke Wakabyashi, A Lawsuit Against Uber
Highlights the Rush to Conquer Driverless Cars, New York Times, Feb.
24, 2017, https://www.nytimes.com/2017/02/24/
technology/anthony-levandowski-waymo-uber-google-lawsuit.html.
A Tesla owner was recently involved in an accident when the
autopilot failed recognize a lane shift in a construction zone,
resulting in a collision with a construction barrier.\10\
---------------------------------------------------------------------------
\10\ Antti Kautonen, Tesla Driver Blames Autopilot for Barrier
Crash, Autoblog, Mar. 3, 2017, http://www.autoblog.com/2017/03/03/
tesla-autopilot-barrier-crash/.
These accidents should alarm the Committee and the public, but they
are only one of myriad issues with autonomous vehicles. Wide-scale
malicious automobile hacking is no longer theoretical.\11\ Although a
full-scale remote car hijacking is certainly a serious risk to car
owners and others,\12\ hijacking is not the only risk posed by
connected car vulnerabilities.\13\ Connected cars leave consumers open
to car theft, data theft, and other forms of attack as well. These
attacks are not speculative; many customers have already suffered due
to vulnerable car systems. For example, criminals have exploited
vulnerabilities in connected cars to perpetrate car ``ransomware''
scams, ``where a car is disabled by malicious code until a ransom is
paid.'' \14\
---------------------------------------------------------------------------
\11\ Brief of Amicus Curiae EPIC, Cahen v. Toyota Motor
Corporation, No. 16-15496 (9th Cir. Aug. 5, 2016), available at https:/
/epic.org/amicus/cahen/EPIC-Amicus-Cahen-Toyota.pdf.
\12\ See, e.g., Andy Greenberg, Hackers Remotely Kill a Jeep On the
Highway--With Me in It, Wired (July 21, 2015), https://www.wired.com/
2015/07/hackers-remotely-kill-jeep-highway/.
\13\ See Bruce Schneier, The Internet of Things Will Turn Large-
Scale Hacks Into Real World Disasters, Motherboard (July 25, 2016),
http://motherboard.vice.com/read/the-internet-of-
things-will-cause-the-first-ever-large-scale-internet-disaster
(explaining that information systems face three threats: theft (i.e.,
loss of confidentiality), modification (i.e., loss of integrity), and
lack of access (i.e., loss of availability)).
\14\ Nora Young, Your Car Can be Held for Ransom, CBCradio (May 22,
2016), http://www.cbc.ca/radio/spark/321-life-saving-fonts-ransomware-
cars-and-more-1.3584113/your-car
-can-be-held-for-ransom-1.3584114.
---------------------------------------------------------------------------
Car manufacturers must adopt data security measures. Early
mitigation of threats to public safety may reduce auto fatalities, spur
innovation, and result in safer vehicles.\15\ There should be great
concern that each of autonomous car maker wants to be the first to have
their vehicle available to the public can poses substantial safety
risks.\16\ A functioning autonomous vehicle does not mean security and
the race to be the first with a functioning, marketable autonomous
vehicle jeopardizes the safety and security of consumers.
---------------------------------------------------------------------------
\15\ See generally, Ralph Nader, Unsafe at Any Speed (1965).
\16\ Mike Isaac, Lyft and Waymo Reach Deal to Collaborate on Self-
Driving Cars, New York Times, May 14, 2017, https://www.nytimes.com/
2017/05/14/technology/lyft-waymo-self-dri
ving-
cars.html?rref=collection%2Fsectioncollection%2Ftechnology&action=click&
contentCollection
=technology®ion=stream&module=stream_unit&version=latest&contentPlace
ment=3&pgtype
=sectionfront; Alex Davies, Detroit Is Stomping Silicon Valley in the
Self-Driving Car Race, Wired, Apr. 3, 2017, https://www.wired.com/2017/
04/detroit-stomping-silicon-valley-self-driving-car-race/.
---------------------------------------------------------------------------
Recently, Charlie Miller, whose research led Chrysler to recall 1.4
million vehicles after he hacked into a Jeep,\17\ stated the danger in
self-driving ridesharing and taxi services stating that ``Autonomous
vehicles are at the apex of all the terrible things that can go wrong .
. . Cars are already insecure, and you're adding a bunch of sensors and
computers that are controlling them . . . If a bad guy gets control of
that, it's going to be even worse.'' \18\ The potential risks that
connected cars pose to the driver, as well as the potential risk to the
public, cannot be understated.
---------------------------------------------------------------------------
\17\ Andy Greenberg, After Jeep Hack, Chrysler Recalls 1.4 Million
Vehicles for Bug Fix, Wired, Jul 24, 2015, https://www.wired.com/2015/
07/jeep-hack-chrysler-recalls-1-4m-vehicles-bug-fix/; Andy Greenberg,
Hackers Remotely Kill A Jeep on the Highway--With Me In It, Wired, Jul.
21, 2015, https://www.wired.com/2015/07/hackers-remotely-kill-jeep-
highway/; Andy Greenberg, The Jeep Hackers Are Back To Prove Car
Hacking Can Get Much Worse, Wired, Aug. 1, 2016, https://www.wired.com/
2016/08/jeep-hackers-return-high-speed-steering-acceleration-hacks/.
\18\ Andy Greenberg, Securing Driverless Cars From Hackers Is Hard.
Ask The Ex-Uber Guy Who Protects Them, Wired, Apr. 12, 2017, https://
www.wired.com/2017/04/ubers-former-top-hacker-securing-autonomous-cars-
really-hard-problem/.
---------------------------------------------------------------------------
In paving the way for the development and deployment of self-
driving cars this Committee should be aware of the National Highway
Traffic Safety Administration's recent proposals that would pre-empt
states from being involved in this process. Historically, Federal
privacy laws have not preempted strong state protections or enforcement
mechanisms.\19\
---------------------------------------------------------------------------
\19\ See e.g., Electronic Communications Privacy Act; Right to
Financial Privacy Act; Cable Communications Privacy Act; Video Privacy
Protection Act; Employee Polygraph Protection Act; Telephone Consumer
Protection Act; Driver's Privacy Protection Act; Gramm-Leach-Bliley
Act.
---------------------------------------------------------------------------
However, NHTSA recently proposed issuing a new Federal Motor
Vehicle Safety Standard for vehicle-to-vehicle communications \20\ and
the Federal Automated Vehicle Policy envisions NHTSA issuing FMVSS's as
connected cars are developed.\21\ However, under the Vehicle Safety Act
the states are pre-empted from issuing any standards for vehicle
performance if it is not identical to an existing FMVSS that regulates
the same aspect of performance.\22\ As EPIC recently explained to the
NHTSA:
---------------------------------------------------------------------------
\20\ Federal Motor Vehicle Safety Standards; V2V Communications, 82
Fed. Reg. 3,854 (Jan. 12, 2017).
\21\ Nat'l Highway Traffic Safety Admin., Federal Automated
Vehicles Policy (Sep. 2016).
\22\ ``When a motor vehicle safety standard is in effect under this
chapter, a State or a political subdivision of a State may prescribe or
continue in effect a standard applicable to the same aspect of
performance of a motor vehicle or motor vehicle equipment only if the
standard is identical to the standard prescribed under this chapter.''
49 U.S.C. Sec. 30102(b)(1).
---------------------------------------------------------------------------
States have a unique perspective allowing them to develop
innovative programs to protect consumers. As [connected car]
technology evolves, the states should be allowed to operate as
laboratories of democracy--a role they have traditionally held
in the field of privacy.\23\
---------------------------------------------------------------------------
\23\ V2V comments at 10.
---------------------------------------------------------------------------
The Committee must make clear that the states must continue to have
the same pivotal role for developing privacy regulations that they have
traditionally held.
EPIC urges this Committee to take these accidents and security
flaws into account as it examines the future of transportation as it
relates to these vehicles. In addition to the substantial privacy
concerns that connected cars present,\24\ these recent incidents show
that there are substantial safety concerns to everyone on the road.
National minimum standards for safety and privacy are needed to ensure
the safe deployment of connected vehicles.
---------------------------------------------------------------------------
\24\ 8 U.S. Gov. Accountability Office, GAO-14-649T, Consumers'
Location Data: Companies Take Steps to Protect Privacy, but Practices
Are Inconsistent, and Risks May Not be Clear to Consumers (2014),
http://gao.gov/products/GAO-14-649T; Jeff John Roberts, Watch Out That
Your Rental Car Doesn't Steal Your Phone Data, Fortune, Sep. 1, 2016,
http://fortune.com/2016/09/01/rental-cars-data-theft/.
---------------------------------------------------------------------------
Auto manufacturers have a particular responsibility to ensure the
safety of drivers. Mr. Mitch Brainwol from the Alliance of Automobile
Manufacturers should be asked:
What are automobile manufacturers doing to ensure data
security in connected cars?
Mr. Rob Csongor from NVIDIA Corporation should be asked:
What information is NVIDIA collecting on drivers?
What is NVIDIA doing to secure driver data?
What cybersecurity measures has NVIDIA adopted to minimize
the risk that its cars can be hacked?
We ask that this Statement be entered in the hearing record. EPIC
looks forward to working with the Committee on these issues.
Sincerely,
Marc Rotenberg
EPIC President
Caitriona Fitzgerald
EPIC Policy Director
Kim Miller
EPIC Policy Fellow
______
Prepared Statement of John Bozzella, President and CEO,
Association of Global Automakers, Inc.
On behalf of the Association of Global Automakers (``Global
Automakers''), I am pleased to provide the following statement for the
record of the Senate Committee on Commerce, Science, and Transportation
hearing entitled ``Paving the Way for Self-Driving Vehicles.'' We
commend Chairman Thune, Ranking Member Nelson, and Senator Peters for
their commitment to pursue bipartisan legislation on automated vehicles
as reflected in their recently released principles.
Global Automakers represents international automobile manufacturers
that design, build, and sell cars and light trucks in the United
States. Our automaker members have invested $56 billion in U.S.
facilities and directly employ 98,500 employees located throughout the
United States. Global Automakers' members have 28 manufacturing
facilities in twelve states and built 4.6 million vehicles in the
United States in 2016, a 41 percent increase in production in the last
decade. Global Automakers and our member companies are committed to
creating the safest, cleanest and most technologically advanced
vehicles on the road.
The automotive industry is making major investments in the research
and development of automated vehicle technology here in the United
States, and Global Automakers thanks the Committee its interest and
proactive approach to vehicle automation. As you know, the Department
of Transportation (DOT) has demonstrated Federal leadership and
established a process to provide assurance that safety is being
addressed at the national level through the issuance of its Federal
Automated Vehicles Policy. At the same time, a number of states around
the country are pursuing policies to regulate automated vehicles.
Therefore, it is important and timely that the Committee is focusing
its attention on this subject. With the right policies, the United
States can continue to lead in the development of these technologies
and bring their benefits to the American people as quickly as possible.
Automated vehicle systems can save lives, enhance mobility, improve
transportation efficiency, and reduce fuel consumption. Public policy
should spur this innovation, encourage testing, and permit nationwide
deployment of vehicles across all levels of automation. Decisions made
today will determine how fast and how far our systems evolve, and
inconsistent policy approaches--particularly as they relate to vehicle
characterization, performance, and design--could have long-lasting
consequences. Only the Federal Government can ensure a framework that
encourages and enables the development of highly automated vehicles,
and it should work with state and local policymakers to provide
guidance and establish clear policy roles and responsibilities.
The United States has long recognized that the automobile market is
a national market and that manufacturers' success rests on the ability
to sell vehicles that can be operated in all fifty states. The U.S.
further recognizes that vehicle safety is a national priority, and the
Motor Vehicle Safety Act has set clear limits on the role of states in
regulating the design of motor vehicles. As opposed to the regulation
of conventional vehicles comprised primarily of mechanical systems,
automated vehicles have systems that rely as much on software as on
hardware. This presents new challenges for policymakers and regulators.
To address these emerging issues, DOT has established a nimble and
flexible approach to oversee the safe testing and deployment of
automated driving systems. In addition to the Federal Policy, the
National Highway Traffic Safety Administration (NHTSA) maintains
existing authority over motor vehicle safety to exercise recall and
enforcement powers to address areas where there is deemed to be an
unreasonable risk to safety. Unfortunately, while NHTSA examines
potential future regulatory action for automated vehicle systems, the
current scope of the Motor Vehicle Safety Act does not explicitly
prevent states from developing disparate requirements in the absence of
Federal standards.
Global Automakers support efforts by Congress and the
Administration to ensure that there is a consistent national approach
to automated vehicle policy in the United States and clarify the
respective roles of federal, state and local governments. As stated in
the Federal Automated Vehicle Policy, the primary roles and
responsibilities of the Federal Government include setting Federal
Motor Vehicle Safety Standards (FMVSS), ensuring compliance with
standards, investigating defects, and issuing guidance for
manufacturers and other entities. In contrast, state responsibility
focuses on issues related to the operation of those vehicles on their
roads, such as driver licensing and responsibility, vehicle
registration, and insurance. This approach has supported decades of
improvement in motor vehicle safety. Preempting state laws and
regulations that prescribe design and performance standards for
automated vehicles would help spur the further development, testing and
deployment of this life-saving technology.
Congress can also help expedite automated technology deployment by
providing NHTSA with authority to exempt an increased number of highly
automated vehicles from standards that would otherwise limit their
deployment. Existing regulations, understandably, did not envision the
emergence of automated vehicle technology, and as a result there is
uncertainty when seeking to certify a vehicle that is designed to
operate without the engagement, or possibly the presence, of a driver.
As noted in a March 2016 report by the U.S. DOT Volpe Center, there are
a number of FMVSS that may limit the deployment of automated vehicles
due to either explicit or implicit references to the presence of human
driver.
Similarly, while expanded exemptions may provide greater
opportunities for the deployment of automated vehicle systems in the
short-term, it may not provide the necessary long-term certainty for
manufacturers. It is therefore important that the Federal Government
identify any outdated standards that may unnecessarily limit innovation
and work collaboratively with industry and other stakeholders to update
those standards to accommodate automated systems.
Global Automakers and our member companies believe that automated
vehicle technologies can provide significant benefits, and we look
forward to working with the Committee to help bring these benefits to
the American people. If policymakers can ensure an environment where
innovation is permitted to thrive, automated vehicles can truly
transform vehicle transportation.
______
Statement of the National Association of Mutual Insurance Companies
The National Association of Mutual Insurance Companies (NAMIC) is
pleased to provide comments to the Senate Committee on Commerce,
Science, and Transportation as it considers the hurdles for testing
deployment of successful automated driving systems (ADS) deployment and
the roles state and Federal governments will play in promoting
innovation and American competitiveness. We appreciate the Committee's
focus on this important matter with the potential to greatly impact the
domestic U.S. property/casualty insurance industry.
NAMIC is the largest property/casualty insurance trade association
in the country, with more than 1,400 member companies representing 39
percent of the total market. NAMIC supports regional and local mutual
insurance companies on main streets across America and many of the
country's largest national insurers. NAMIC member companies serve more
than 170 million policyholders and write more than $230 billion in
annual premiums. Our members account for 54 percent of homeowners, 43
percent of automobile, and 32 percent of the business insurance
markets.
NAMIC fully supports ADS innovation and competitive development
that enhances safety. In this context, ADS innovation will be the
deliberate application of information, imagination, and initiative by
which new ideas are generated and converted into greater driving
safety. It is important to note, however, that while technology is a
key part of ADS innovation, it is not the only part. Especially in the
case of a revolutionary change in mobility as will occur with ADS,
successful and fruitful outcomes will only materialize when innovative
approaches to cybersecurity, regulation, consumer behavior, and market
practices. Integrating innovative practices, operations, rules, and
regulations will be just as necessary to ADS innovation as technology.
With respect to ADS, NAMIC has participated in National Highway
Traffic Safety Administration (NHTSA) panels relating to state
jurisdiction and pre-market approval, serves as a Board Member of the
Advocates for Highway and Auto Safety, and is working with the
Insurance Institute for Highway Safety supporting the Virginia Tech
Transportation Institute as part of the National Cooperative Highway
Research Program.
Automated Driving Technology
Enthusiastic advocates of ADS contend that its ability to address
driver error is its greatest purpose. NHTSA attributes 94 percent of
all car crashes on human error, and ADS advocates infer or directly
state that ADS will eliminate that human error, and thus the
overwhelming majority of car crashes. That oft-cited 94 percent comes
from NHTSA's ``Critical Reasons for Crashes Investigated in the
National Motor Vehicle Crash Causation Survey'', which reported data
from The National Motor Vehicle Crash Causation Survey conducted from
2005 to 2007. The critical reason, which is the last event in the crash
causal chain, was assigned to the driver in 94 percent (2.2
percent) of the crashes in that survey. NHTSA defines the `critical
reason' as only the immediate reason for the critical pre-crash event,
and simply the last failure in the causal chain of events leading up to
the crash.
Critical reasons were broadly classified into recognition errors,
decision errors, performance errors, and non-performance errors:
Recognition error, which includes driver's inattention,
internal and external distractions, and inadequate
surveillance, was the most frequently assigned critical reason
(41 percent 2.2 percent).
Decision error, such as driving too fast for conditions, too
fast for a curve, false assumption of others' actions, illegal
maneuver and misjudgment of gap or others' speed accounted for
about 33 percent (3.7 percent) of the crashes.
In about 11 percent (2.7 percent) of the
crashes, the critical reason was performance error such as
overcompensation, poor directional control, etc.
Sleep was the most common critical reason among non-
performance errors that accounted for 7 percent
(1.0 percent) of the crashes.
Other driver errors were recorded as critical reasons for
about 8 percent (1.9 percent) of the drivers.
These above definitions are relevant to the proposed application of
ADS as a remedy to these specific errors. The central premise behind
the development and broad application of ADS technology is that ADS
will have fewer of these specific recognition errors, decisions errors,
performance errors and other errors than human drivers represented by
the survey. ADS function that results in significant reductions in
these specific errors should have a corresponding reduction in crashes
and result in greater vehicle and personal safety.
The Hurdles for Testing and Successful ADS Deployment
ADS may have the potential to be much safer than human drivers, and
could result in a dramatic decrease in crashes and highway deaths. But
ADS are also under development, complex, and include numerous known and
unknown hazards as well as unintended consequences. There are literally
millions of related technological and policy questions as well as
questions related to liability, cybersecurity, vehicle connectivity,
and infrastructure.
Possibly the most challenging area is how ADS technology will deal
with the transitional period, when human drivers share the road with
ADS and continue to make the same and new recognition errors, decisions
errors, performance errors and other errors. As policies and
regulations are made toward enabling this ambitious ADS safety
environment, communicating relevant data and analysis is critical to
verify the safest transition from human-controlled to automated
driving.
The technical challenges to ADS testing and successful development
are vast and complex; too vast and complex to even begin to enumerate
here. But even if these thousands of technical challenges can be
successfully addressed, ADS will not be adopted unless consumers can be
convinced that ADS use will be safe and reliable. Recent surveys by
both the Massachusetts Institute of Technology (MIT) and J.D. Power
have indicated that consumer comfort or trust in full automation
appears to actually be declining. A shift away from trust in automation
was observed across all age groups, and most notably in younger age
ranges (which had been most open to ADS in the past). Roughly half of
those surveyed stated that they would never purchase an ADS vehicle.
As consumers learn more about ADS, they appear to have more
questions rather than answers. A critical hurdle for ADS deployment is
that consumers lack relevant information to adequately gauge the
performance and potential benefits of ADS. Consumers will require
objective assessment and evaluation of just how much safer ADS are than
human drivers, and in what conditions. Consumer acceptance will require
expert collection and evaluation of data and analysis on the ADS as
designed, as well as objective data and analysis from crashes involving
ADS.
The history of the auto insurance industry provides ample evidence
of that experience and expertise, as well as the commitment of the
insurance industry to enhancing driving safety. A critical requirement
for the testing and successful development of ADS will be insurers
using their decades of expertise to provide objective data-based safety
evaluations. This will require ADS companies to collect uniform ADS
design and safety information and make that information available to
insurers to enable an objective assessment of ADS safety. Insurers will
also need access to ADS information and data--including crash and
incident information and data--that is timely, complete, and useful.
The insurance industry also supports organizations dedicated to
understanding crash risk and furthering safety enhancements. The
Insurance Institute for Highway Safety is the premier organization
dedicated to reducing the losses--deaths, injuries, and property
damage--from crashes on the Nation's roads. The Highway Loss Data
Institute provides scientific studies of insurance data representing
the human and economic losses resulting from the ownership and
operation of different types of vehicles and by publishing insurance
loss results by vehicle make and model. Advocates for Highway and Auto
Safety is an alliance of safety groups and insurance companies working
together to make America's roads safer.
State and Federal Roles to Ensure Safety While Promoting Innovation and
American Competitiveness
Promoting innovation and competitiveness will require joint
development of state and Federal rules and regulations. NAMIC advocates
that the roles of states and Federal authorities would best facilitate
ADS safety development as follows:
1. The Federal Government--through NHTSA--should have the authority
to make determinations for the required performance and safety,
as well as data integrity, of ADS.
2. States and localities should have the authority to make the
determinations of the registration, licensing, and operation of
ADS in that state/locality.
3. States should retain the regulation of ADS insurance for the
vehicle and/or operator.
4. States should define and address ADS personal liability issues in
state/tort law and regulation in line with existing liability
constructs. States and Federal authorities should have the
authority to define and address ADS liability issues in law and
regulation.
5. States and Federal authorities working together should make clear
and workable data security and privacy requirements for
vehicles with ADS.
In summary, NAMIC supports ADS development, and insurers are
leading advocates toward 100 percent adoption of safe ADS. The
realistic support of the potential for greater safety requires that the
business of insurance put forth significant effort to objectively
identify and analyze facts and data that comes naturally in the
business of insurance. The development and adoption of ADS will require
substantial policy and structure changes, which will also require the
same factual analysis and review. As the primary rationale for ADS
development is reducing vehicle crashes, injuries and fatalities, it is
important that insurers have an active role in the assessment and
communication of those benefits.
NAMIC greatly appreciates the opportunity to provide this testimony
to the Committee and looks forward to working with the Committee in the
development of ADS policy and regulation.
______
Consumer Watchdog
June 14, 2017
Sen. John Thune, Chairman
Sen. Bill Nelson, Ranking Member
Senate Committee on Commerce, Science, and Transportation
Washington, DC.
Re: Hearing: ``Paving the Way for Self-Driving Vehicles''
Dear Chairman Thune and Ranking Member Nelson,
We are writing on behalf of Consumer Watchdog, a national
nonpartisan, nonprofit public interest group to submit our just-issued
report, ``Self-Driving Vehicles: The Threat to Consumers,'' into the
formal written record of the Commerce Committee's public hearing titled
``Paving the Way for Self-driving Vehicles.''
We must also express our deep dismay at the Committee's failure to
include any representatives of consumer groups among the witnesses
called to speak at the hearing. The witness panel includes a spokesman
for auto manufacturers, a representative of a company developing robot
car technology, and organization developing a test center for robot car
technology. A spokesman for MADD has the laudable, but narrow, agenda
of combating drunk driving. This is an industry-dominated panel with no
representatives of auto safety or consumer protection organizations.
As our report shows, robot cars operating without mandatory safety,
security, privacy and ethical standards will pose unprecedented risks
to the American public. Lost in the hyperbole over robot cars is a
realistic assessment of the likely costs to both consumers and
taxpayers particularly over the coming decades, when robot cars and
human drivers will share a ``hybrid highway.''
That period will feature complex interactions between people,
computers, cars and public streets and freeways, during which today's
liability protections and rules restraining insurance rates and unfair
practices by insurance companies will become especially important. To
deal with the challenge posed by autonomous vehicle technology,
Consumer Watchdog believes six principles must be adopted. They are:
1. Protect the civil justice system. The state-based civil
justice system--open courts, impartial judges and citizen
juries--is fully equipped to handle the determination of legal
responsibility as our transportation system evolves over the
coming decades. Disputes over who is at fault in a crash
involving a self-driving car or truck will require the full
power of civil justice system, with its procedural safeguards
of an impartial judge, full public transparency, and trial by
citizen juries, to investigate and publicly expose the cause of
crashes, compensate the victims for deaths, injuries and
property damage, punish the wrongdoer, and force manufacturers
to make changes in their products to prevent future harm. When
their autonomous technologies fail, hardware and software
manufacturers must be held strictly liable. Lawmakers should
reject legislation to limit or restrict state consumer
protection laws. Manufacturers must not be permitted to evade
these consumer protections by inserting arbitration clauses,
``hold harmless'' provisions or other waivers in their
contracts.
2. Enact stronger state consumer protections against insurance
company abuses. According to a 2013 report by the Consumer
Federation of America, ``California stands out from all other
states in having the best insurance regulatory system for
protecting consumers.'' Enacted by California voters in 1988,
California's insurance reform law provides precisely the
stronger protections consumers will require in the era of robot
vehicles. The reforms, known as Proposition 103, have protected
motorists (along with homeowners, renters, businesses and
medical providers) against unjust insurance rates (including
product liability insurance rates) and anti-consumer and
discriminatory practices. The law's emphasis on rewarding
drivers with lower insurance premiums based on their safety
record, their annual mileage, their driving experience, and
other rating factors within their control that are
``substantially related to the risk of loss,'' will be critical
in the new automotive era. Proposition 103's mandate for public
disclosure and public participation in regulatory matters are
essential components of a system that will be trusted by
consumers.
3. Enact auto safety standards. Private companies cannot be
trusted to develop and deploy robot cars and trucks without
rules. The Federal auto safety agency or other relevant Federal
agencies, or in their absence, state auto safety agencies, must
develop standards for the testing and deployment of the
multiple technologies required by robot vehicles. These
standards must address safety; security; privacy and the
software that determines the robot's actions in the event of an
impending collision and as it makes life and death decisions.
They must be enforceable by consumers in courts of law.
4. Stronger laws are needed to protect consumers' privacy. The
laws have not kept pace with the evolution of technology and
the collection and monetization of consumers' personal data.
Hardware and software manufacturers and insurance companies
must be barred from utilizing tracking, sensor or
communications data, or transferring it to third parties for
commercial gain, absent separate written consent (which should
not be required as a condition of accessing the services of the
vehicle/manufacturer, and which should be revocable by the
consumer at any time).
5. Bar Federal interference in state consumer protection laws.
Neither Congress not Federal agencies should be permitted to
preempt or override stronger state based civil justice,
insurance reform or auto safety laws.
6. Respect democratic and human values. The sponsors of self-
driving vehicles have promoted the myth that machines are
infallible in order to justify the wholesale departure from a
panoply of norms that form founding principles for the nation,
beginning with the rule of law; individual and corporate
responsibility; long held legal principles that distinguish
between human beings and property; and the transparency of
public officials and institutions that is a hallmark of
democracy. The strategy of substituting robot values for human
values has reached its apotheosis in the determination by robot
car company executives to program computers to make life and
death decisions, and to keep that decision-making process
secret. Lawmakers will need to impose the rule of law and other
attributes of American democracy upon the executives of the
hardware and software companies that manufacture self-driving
cars.
In conclusion, Congress must not succumb to the siren song of the
autonomous car developers who are over promising what autonomous
vehicle technology can do today. We call on you to require the
development of enforceable Federal safety performance standards.
Responsible regulation goes hand-in-hand with innovation. Voluntary
``standards'' in the auto industry have repeatedly been proven to be
weak and insufficient. Safety must come before the automakers' bottom
lines. Consumer Watchdog calls on you to enact the necessary
regulations to protect the safety of our highways.
Sincerely,
Harvey Rosenfield John M. Simpson Joan Claybrook
Founder Privacy Project Former President Public
Director Citizen and Former NHTSA
Administrator
______
Self Driving Vehicles
The Threat to Consumers
By Harvey Rosenfield
Table of Contents
1.0 Introduction
2.0 Robot Cars Will Pose Unprecedented Safety, Security, and Privacy
Risks
2.1 Risk: Defective Hardware and Buggy or Biased Software
2.2 Risk: Privacy and Security
2.3 Risk: Regulatory Failure
2.4 Current Status of Robot Cars
2.5 Current Status of Intelligent Transportation Infrastructure
2.6 The ``Driverless Divide''
2.7 The ``Hybrid Highway''
3.0 The Personal Responsibility System and Self-Driving Vehicles
3.1 Tort liability
3.1.1 Negligence
3.1.2 Product liability
3.1.3 Common Carrier Liability
3.1.4 Liability Scenarios
TABLE: Legal System--Risk/Liability Matrix
3.2 Insurance
3.2.1 Insurance and autonomous vehicles
3.2.2 The Proposition 103 Model
4.0 The Industry Agenda to Roll Back Consumer Rights
4.1 Restrictions on liability laws to encourage ``innovation''
4.2 Restrictions on liability laws to ``lower insurance rates''
4.3 Repealing protections against insurance company price gouging and
discrimination
5.0 Guiding Principles
______
1.0 Introduction
Self-driving vehicles have become a cultural and political
phenomenon. To peruse the breathless headlines is, like a ride in Marty
McFly's DeLorean, to experience the sensation of visiting a wondrous
future. Millions of hours previously wasted in traffic, searching for a
parking spot or waiting in line at the DMV, will be restored to us,
while we sit back in our passenger seats, dine, work or watch movies as
robots whisk us around. It's a world in which there are virtually no
car accidents, because infallible computers will replace impatient,
emotional, tired, distracted, all too error-prone human beings.
And, wondrously, there will be no need to write another check to
the auto insurance company. In the highly unlikely event that anything
goes wrong, the robot car manufacturers will stand behind their
products and pick up the tab, no questions asked.
Not so fast.
America is decades away from a completely ``self-driving''
transportation system. But the insurance and auto industries are
already preparing to exploit the prospect of self-driving robot cars
and trucks so they can force Americans to pay more for insurance on the
cars they own or lease, and roll back state consumer protection laws so
that when their self-driving vehicle gets in a crash, it will always be
the consumer's fault.
The sparkly chimera of robots replacing human drivers--freeing
people to spend their drive time more enjoyably and productively--has
captivated the public and media, driven by self-interested auto
manufacturers and software developers. But there has been very little
public discussion of whether self-driving vehicles will coexist or
collide with long-standing principles of accountability, transparency,
and consumer protection that collectively constitute the Personal
Responsibility System.
The Personal Responsibility System is a set of state-based
liability and insurance laws that dispenses justice: First, by
regulating insurance companies to mandate fair auto insurance premiums
and rate setting practices that emphasize a motorist's safety record
rather than surrogates for wealth, race or creed. Second, by
encouraging the safe manufacture, marketing and operation of cars and
trucks. Third, by determining fault and compensation for, deaths,
injuries and property damage caused by defective cars or the negligent
operation of vehicles. Fourth, by making sure that auto insurance
companies handle crash claims fairly by paying promptly and fully.
At the intersection of the Personal Responsibility System,
insurance laws and robot cars, California is of particular interest.
Sweeping reforms passed by the voters in November 1988, known as
Proposition 103, have protected motorists (along with homeowners,
renters, businesses and medical providers) against excessive rates, and
discriminatory practices by insurance companies that have historically
targeted individuals they deem ``undesirable'' based on their race,
religion, or income. Thanks to Prop 103, California is the only state
in the Nation where the average auto insurance premium went down
between 1989 and 2010, according to a report by the Consumer Federation
of America, saving motorists alone over $100 billion in premiums since
1989;\1\ CFA's report concluded that ``California stands out from all
other states in having the best regulatory system for protecting
consumers.''
---------------------------------------------------------------------------
\1\ ``What Works? A Review of Auto Insurance Rate Regulation in
America,'' Consumer Federation of America, November 12, 2013.
---------------------------------------------------------------------------
Dating back to the American Revolution and enshrined in the Bill of
Rights, the Personal Responsibility System has made the American
marketplace a paragon of safety, fairness and prosperity.
But insurance companies and automakers now say that it's outdated
and incompatible with self-driving vehicles.
They argue that America can dispense with the civil justice
system--open courts, impartial judges and citizen juries--because these
core consumer protections will ``chill this promising technology
[autonomous vehicles] and the huge advances in overall public safety it
promises,'' as the leading lobbying group for corporate defendants
recently put it.\2\ That the manufacturing and insurance industries are
exploring ways in which they can limit or shift their responsibility is
not particularly surprising, given that safety-related costs and claims
are likely to increase as the result of the new, riskier and so-far
unregulated technologies. However, these strategies--historically
employed by industries seeking a government bailout of risks--undermine
competition and distort incentives in the marketplace.
---------------------------------------------------------------------------
\2\ U.S. Chamber of Commerce Institute for Legal Reform, ``Torts of
the Future: Addressing the Liability and Regulatory Implications of
Emerging Technologies,'' March 2017, p.2
---------------------------------------------------------------------------
The insurance industry is also opportunistically targeting consumer
protections against insurance company abuses like overcharging
consumers and discrimination, claiming they will be unnecessary once
self-driving vehicles arrive. In California, insurance companies are
using robot cars as the excuse to challenge Proposition 103. As a
recent report explained, ``Over three decades, insurance companies have
spent millions of dollars trying to chip away at Prop 103's regulations
both through litigation and at the ballot box--with little success.
Now, however, the industry has found a new source of optimism in a
different phrase: driverless cars.'' \3\
---------------------------------------------------------------------------
\3\ C. Miller, ``Driverless Cars Give Insurers New Vehicle to
Criticize California's Rates Law,'' The Recorder, May 19, 2017 (http://
www.therecorder.com/home/id=1202786779265/Driverless-Cars-Give-
Insurers-New-Vehicle-to-Criticize-Californias-Rates-
Law?mcode=1202617072607&cur
index=0).
---------------------------------------------------------------------------
And if history repeats itself, the insurance and auto lobbies may
ask the Trump Administration to impose Federal rules that would
override the Personal Responsibility Laws of the fifty states.
Even a cursory analysis of the risks that robot cars and trucks
will pose over the coming years shows that the industries' argument is
wrong. Issues of corporate responsibility, liability and insurance will
become far more important as self-driving vehicles are rolled out.
To understand the crucial role that the Personal Responsibility
System will play in the coming decades, two points are critical.
First, a fully autonomous transportation system is decades away at
best. No completely self-driving vehicle is offered for sale today, and
notwithstanding a great deal of marketing hype, no manufacturer has set
a firm date when it will market a passenger vehicle that is able to
operate in all conditions without human intervention, or, importantly,
what it will cost to buy.\4\ Indeed, the system of vehicle-to-vehicle,
vehicle-to-satellite, and vehicle-to-road sensor communications
infrastructure that would enable tens of millions of vehicles to
simultaneously, securely and autonomously operate in proximity to each
other on streets and highways without human intervention is barely in
the planning stages. Nor is there any consensus on how local, state and
Federal governments will pay for it. After all, most municipalities
these days are struggling to fill potholes. And it is far from clear
that every American consumer is going to be ready to abandon America's
love affair with the open highway, or to surrender the steering wheel
to a machine that is going to cost many thousands of dollars more than
today's vehicles.
---------------------------------------------------------------------------
\4\ A number of automakers have proclaimed they will sell
autonomous vehicles over the next few years, but are short on the
specifics. Ford has announced it intends to have a ``fully autonomous''
vehicle for commercial ride-sharing or ride-hailing applications by
2021, but according to the fine print the vehicle will offer only
``high,'' not ``full,'' automation. (https:// media.ford.com/content/
fordmedia/fna/us/en/news/2016/08/16/ford-targets-fully-autonomous-
vehicle-for-ride-sharing-in-2021.html.)
---------------------------------------------------------------------------
Even if we assume that someday fully autonomous vehicles will be
safe enough to deploy, and that all Americans will be ready and able to
surrender the steering wheel,\5\ for the foreseeable future traditional
vehicles driven by humans will share a ``hybrid highway'' filled with
cars and trucks of widely varying degrees of automation and autonomy.
Relatively few of them will be truly self-driving.
---------------------------------------------------------------------------
\5\ It is more likely that self-driving vehicles, when they become
available, will be initially adopted by commercial enterprises such as
ride-sharing operations.
---------------------------------------------------------------------------
Second, the argument that robot cars and trucks will virtually
eliminate crashes is based on a fallacy: that machines are infallible.
It makes sense that carefully tested automation technologies will
improve the safety of cars and trucks in the future. However,
completely self-driving cars don't exist yet and we don't know how they
will change transportation patterns once they arrive. So for the
moment, the claim that robot vehicles will dramatically reduce
vehicular deaths, injuries and property damage is simply speculation.
But we know this: machines make mistakes--sometimes catastrophic
mistakes. Consider the automation-related mass disasters that have
befallen the commercial airline industry in recent years,
notwithstanding its self-avowed goal of zero tolerance for failure.
Google/Waymo boasts that its computer-controlled test vehicles have
logged the equivalent of over 300 years of human driving experience.\6\
But the duration of testing that would be required in order to match
the safety tolerance of commercial airplanes is 114,000 years.\7\
---------------------------------------------------------------------------
\6\ See https://x.company/waymo/ (last visited June 12, 2017).
\7\ See B.W. Smith, Automated Driving & Product Liability, 2017
Mich. St. L. Rev. 1, at 31 re ``at least a billion hours of testing,''
also citing Philip Koopman & Michael Wagner, Challenges in Autonomous
Vehicle Testing and Validation, 4 SAE Int'l. J. Transp. Safety 15, 15-
16 (2016).
---------------------------------------------------------------------------
In any case, even if robot cars and trucks someday become 100
percent safe, we can say with certainty that in the short term,
autonomous vehicles will pose new and unprecedented risks as they
interact with traditional cars and trucks on the hybrid highway.
The #1 safety threat posed by self-driving vehicles is bugs or
biases built into the robots' brains.
A crucial and controversial component of the self-driving car or
truck is the set of algorithms that will determine how the vehicle
responds when confronted with an unexpected, life-threatening emergency
such as children playing in the street, pedestrians, roadside
construction, and weather conditions. Initially these rules will be
programmed by corporate engineers; eventually the engineers will teach
the cars how to think for themselves using artificial intelligence, so-
called ``machine learning.''
These algorithms will be responsible for life and death decisions
that will place their financial interests in conflict with their
customers' lives. But Google and other software developers have refused
to disclose to the public or regulators the robot values that they are
programming into their computers to replace human values and judgment.
When Google's self-driving vehicle sideswiped a bus in Mountain View,
California, the company called it a ``misunderstanding'' between the
bus driver and the robot.\8\ A software ``misunderstanding,'' even at 2
m.p.h., cannot be dismissed. Just as occurs every day on our roadways,
the robot will confront situations in which the choice is not whether
to smash into someone, but rather who to hit--an oncoming vehicle, a
pedestrian in a crosswalk, a mom pushing her infant in a stroller on
the sidewalk?
---------------------------------------------------------------------------
\8\ M. McFarland, ``For The First Time, Google's Self-Driving Car
Takes Some Blame For A Crash,'' Washington Post, February 29, 2016
(https://www.washingtonpost.com/news/innovations/wp/2016/02/29/for-the-
first-time-googles-self-driving-car-takes-some-blame-for-a-crash/
?utm_term=.f19f6bdc6f4d).
---------------------------------------------------------------------------
Other risks include failures in the extremely complex hardware
(Google and other companies' robot test vehicles have been involved in
multiple accidents and hundreds of near-misses \9\); privacy breaches
(now endemic in the United States \10\); criminal hacking or even
terrorist cyber attacks involving hundreds or thousands of vehicles, as
the FBI has warned.\11\
---------------------------------------------------------------------------
\9\ See Section 2.4.
\10\ See Section 2.2.
\11\ See Section 2.2.
---------------------------------------------------------------------------
When one or more of these serious risks inevitably results in a
crash on the ``hybrid highway,'' the inquiry into what caused the crash
and who is responsible will include the manufacturers of the automated
vehicle's hardware and software. There is no reason to believe that
they or their vendors will respond any differently than they do today:
deny their liability and attempt to shift the blame to the human
driver. Indeed, in the limited experience so far, the companies that
have deployed robot technologies have not readily accepted
responsibility for their crashes and near-misses. This is particularly
true of Tesla, which has denied responsibility for two fatalities
involving its ``Auto Pilot'' software.\12\ While some car companies
have stated that they will assume liability for the failure of their
robot technologies, nothing's in writing, and their pledge appears to
be conditioned on a determination that their technology was at
fault.\13\
---------------------------------------------------------------------------
\12\ See Section 2.4.
\13\ See Section 3.1.4.
---------------------------------------------------------------------------
Crashes aside, consumers who own or lease self-driving vehicles
will face far greater responsibility for vehicle maintenance than they
bear today. Self-driving vehicles will be extraordinarily reliant on
external sensors--the eyes and ears of the robot car's brain. An
accidental driveway ding in a sensor could have deadly consequences
when the vehicle is on the road. What happens if the consumer fails to
download a software update? No one has suggested that manufacturers
will be prepared to assume liability for a crash caused by the
consumer's failure to maintain the vehicle's operational status.
That leads us right back to the Personal Responsibility System of
insurance and liability laws.
With the heightened risks that the new automated technologies will
pose over the coming years, the legal requirement that manufacturers be
held strictly liable for defective products, a mainstay of America's
consumer protection regime, will remain essential. Disputes over fault
will require the full power of the civil justice system, with its
procedural safeguards of an impartial judge, full public transparency,
and trial by citizen juries, to investigate and publicly expose the
cause of crashes, compensate the victims for deaths, injuries and
property damage, punish the wrongdoer, and force manufacturers to make
changes in their products to prevent future harm.
For the same reasons, protections against abusive practices by
insurance companies will be critical. So long as motorists face legal
responsibility for the proper maintenance and safe operation of
vehicles they own, lease, rent or control, they will require the same
liability insurance coverage that they purchase today. Pointing to the
new risks of the Hybrid Highway, and the greater costs of repairing the
automation technology, insurance companies will no doubt ask their
customers to pay more for it. Preventing insurance companies from
overcharging motorists, and from high-tech forms of redlining that rely
on manipulation of data about each customer, will necessitate forceful
consumer protections such as those contained in Proposition 103.
The proponents of autonomous vehicles like to describe themselves
as ``disruptors,'' and take pride in refusing to accept the norms of
what they deride as the decrepit status quo. But behind the scenes,
industry players are employing decidedly Old World lobbying and
political strategies to avoid public and regulatory scrutiny and
oversight, while at the same time urging lawmakers to pass legislation
that would limit or even eliminate their legal accountability to
injured consumers.
Consumer Watchdog, a non-profit founded in 1985, has deep roots
working for the public interest on the issues that will be of
tremendous concern to consumers as automated vehicles evolve: civil
justice and corporate accountability; public safety; the premiums and
underwriting practices of the insurance industry; the diminishing
privacy of consumers in the digital age; and the importance of
government oversight, public scrutiny and participation in decision-
making.
This report will discuss the safety, security and other risks posed
by robot cars; why the consumer protections of the Personal
Responsibility System will be critical in the coming decades as self-
driving vehicles come ``on line''; and the campaign by insurance
companies, automakers and possibly even the Federal Government, already
underway, to undermine those essential consumer protections.
2.0 Robot Cars Will Pose Unprecedented Safety, Security, and Privacy
Risks
The safety of driverless vehicles should be the paramount concern
of the auto and insurance industries, if for no other reason than flaws
and failures in automated vehicle systems will impose potentially
enormous, even catastrophic liability upon hardware and software
manufacturers in the event their products cause harm, and lead to more,
and more costly, insurance claims.
In this context, it's worthwhile to consider the current state of
American vehicle safety. Car crashes in the United States killed 35,092
and injured over 2.44 million people in 2015;\14\ including property
damage, the total estimated economic cost of car crashes was estimated
at $242 billion in 2010.\15\ There were a record 801 separate recalls
involving 63.7 million vehicles in 2014, and 613 recalls of 40 million
vehicles as of mid-2015.\16\ Three of the largest recalls in recent
years concern vehicle safety failures--defective GM ignition
switches,\17\ exploding Takata airbags,\18\ and unintended acceleration
in Toyota vehicles \19\--that have taken hundreds of American lives.
---------------------------------------------------------------------------
\14\ NHTSA, 2015 Motor Vehicle Crashes: Overview (August 2016).
\15\ NHTSA, The Economic and Societal Impact Of Motor Vehicle
Crashes, 2010 (Revised May 2015).
\16\ NHTSA, 2014 Vehicle Recalls by Manufacturer (available at
http://www.safercar.gov/Vehicle+Owners/vehicle-recalls-historic-recap);
2015 data: Statement Of Joan Claybrook Consumer Co-Chair Advocates For
Highway And Auto Safety, on ``Examining Ways To Improve Vehicle And
Roadway Safety'' before the Committee On Energy And Commerce
Subcommittee On Commerce, Manufacturing And Trade, October 15, 2015, p.
2.
\17\ ``NHTSA Admits Faults In GM Investigation,'' Detroit News,
June 5, 2015 (http://www.detroitnews.com/story/business/ autos/general-
motors/2015/06/05/gm-nhtsa-report/2854
0239/); ``Why General Motors' $900 Million Fine For A Deadly Defect Is
Just A Slap On The Wrist,'' Washington Post, September 17, 2015
(https://www.washingtonpost.com/news/ business/wp/2015/09/17/why-
general-motors-900-million-fine-for-a-deadly-defect-is-just-a-slap-on-
the-wrist/).
\18\ ``Everything You Need to Know about the Takata Airbag
Recall,'' Consumer Reports, December 23, 2015 (http://
www.consumerreports.org/cro/news/2014/10/everything-you-need-to-know-
about-the-takata-air-bag-recall/index.htm?loginMethod=auto).
\19\ ``Toyota Says It's Settled 338 Cases So Far In Acceleration
MDL,'' Law360, July 22, 2015 (http://www.law360.com/articles/681915/
toyota-says-it-s-settled-338-cases-so-far-in-acceleration-mdl).
---------------------------------------------------------------------------
The unprecedented number of recalls in recent years suggests a
dangerously cavalier attitude toward public safety on the part of
vehicle manufacturers. It raises serious concerns as to whether
manufacturers are presently, or will be, capable of building safe robot
cars and trucks, which will far exceed the complexity and
sophistication of today's vehicles.
When assessing whether autonomous vehicles will ever be 100 percent
safe, consider the transportation sector in which automation is by far
the most advanced and the concern for safety is arguably greater than
in any other: commercial air travel. Recent airline disasters cast
doubt on whether one hundred percent reliance on ``fly-by-wire''
technology will ever be safe.\20\
---------------------------------------------------------------------------
\20\ The crash of Asiana Flight 214 at San Francisco airport in
2013 killed two passengers and injured 181 others; investigators have
determined that the pilots did not understand the highly automated
flight systems and were unable to recover control of the plane when a
crash was imminent. (M. Wald, ``Pilots in Crash Were Confused About
Control Systems, Experts Say,'' New York Times, December 11, 2013.) An
Air France jetliner disappeared into the Atlantic off the coast of
South America in 2009, killing 216 passengers and a crew of twelve,
including three pilots; again, the black box revealed that the pilots
did not understand the plane's automated functions, some of which had
failed. (See W. Langewiesche, ``The Human Factor,'' Vanity Fair
(October, 2014) (http://www.vanityfair.com/business/2014/10/air-france-
flight-447-crash)).
---------------------------------------------------------------------------
To our knowledge, no one has suggested that the manufacturers of
robot cars can or will aim for that level of safety; the former head of
NHTSA suggested in 2016 that autonomous vehicles will merely be twice
as safe as human-driven cars.\21\ By definition, that leaves a lot of
carnage on America's streets.
---------------------------------------------------------------------------
\21\ K. Naughton, ``Regulator Says Self-Driving Cars Must Be Twice
as Safe,'' Reuters, June 8, 2016 (https://www.bloomberg.com/news/
articles/2016-06-08/u-s-auto-regulator-says-self-driving-cars-must-be-
twice-as-safe.)
---------------------------------------------------------------------------
No one disputes that the evolution of motor vehicle technology has
the potential to prevent deaths, injuries and property damage. New
technologies such as automatic emergency braking, lane keeping,
collision warning, and assisted parking are already doing so, and
indeed should be made standard equipment in all vehicles. The point is
that the gradual automation of driving will introduce a new set of
risks. These risks will necessarily be far broader than those posed by
vehicles today--suggesting that the ramifications for liability and
insurance will be significant. A fully autonomous robot-based
transportation system will likely reduce the number of crashes caused
by human error, but that does not tell us anything about the overall
impact of a fully autonomous system.\22\
---------------------------------------------------------------------------
\22\ See B. W. Smith, Automated Driving & Product Liability 2017
Michigan State Law Review 1 for a careful but admittedly provisional
analysis suggesting simplistic assumptions about costs and savings may
be incorrect.
---------------------------------------------------------------------------
2.1 Risk: Defective Hardware and Buggy or Biased Software
The core hardware components of an autonomous vehicle are the
computer and sensors. An array of electronic devices--presently, radar,
lidar, and video--will be responsible for detecting the external
conditions that the vehicle must navigate: road signs, and in their
absence street markings; other vehicles (including bicycles,
motorcycles, trucks); pedestrians (including seniors, and children);
pets; traffic lights; street and highway signs; road construction; law
enforcement activities; weather (fog, snow or heavy rainfall) and other
natural phenomena (such as trees). The on-board computer system will
collect the data from these inputs as well as external communications
sources such as other vehicles, intelligent highways (more on that
below) or satellite based traffic control systems, process the
information and make decisions in a few milliseconds (a millisecond is
1/1000 of a second) that are presently executed by humans. Scientists
have estimated that the human brain is thirty times faster than the
fastest super computer;\23\ it takes a neuron 0.5 milliseconds to
transmit a signal \24\ and 13 milliseconds for the brain to process an
image.\25\ A truly driverless vehicle must be capable of fully
replicating and processing the immense data stream currently processed
by the human brain, such as hand gestures, the facial expressions of
other motorists and pedestrians, and a virtually infinite number of
other variables in the interior and external driving environment.
---------------------------------------------------------------------------
\23\ H. Hsu, ``Estimate: Human Brain 30 Times Faster than Best
Supercomputers,'' IEEE Spectrum, August 26, 2015 (http://
spectrum.ieee.org/tech-talk/computing/networks/estimate-human
-brain-30-times-faster-than-best-supercomputers.)
\24\ http://news.mit.edu/2014/in-the-blink-of-an-eye-0116
\25\ http://www.makeuseof.com/tag/geeks-weigh-in-does-a-human-
think-faster-than-a-computer/
---------------------------------------------------------------------------
Software will analyze the sensor and communications data flow and
instruct the vehicle how to navigate. A particularly critical function
of the software will be to replace the judgment of human motorists not
just to avoid collisions but also to comply with traffic laws and
rules. It will therefore be necessary for the software in driverless
vehicles to make the split-second life and death decisions that human
drivers make today when a collision is unavoidable. Confronted with the
prospect of imminent harm to passengers in other vehicles, pedestrians,
or the occupants of the AV itself, how will the self-driving software
decide which course to take? On what basis will the software make such
decisions? Who will it kill?
These life and death deciding programs will be coded by human
engineers working for private corporations, at least initially.
Will engineers program their computers with human ethics as well as
a database of traffic laws? When Mercedes announced that its software
would protect the occupants of Mercedes vehicles at the expense of
everyone else, it provoked a public firestorm that led Mercedes to
amend its statement.\26\
---------------------------------------------------------------------------
\26\ ``Self-Driving Mercedes-Benzes Will Prioritize Occupant Safety
over Pedestrians,'' Car and Driver Blog (http://blog.caranddriver.com/
self-driving-mercedes-will-prioritize-occupant-safety-over-pedestrians/
).
---------------------------------------------------------------------------
Will companies like Google, that have developed real-time facial
recognition software, write algorithms to avoid harm to high net worth
individuals, thus limiting their own liability? We simply do not know,
because there are no rules that specify the answers to these questions,
and software companies like Google consider their algorithms highly
proprietary and steadfastly refuse to disclose the decision-making
principles, values or formulae that determine the vehicle's actions.
Eventually, the engineers will teach the robot cars and trucks how
to learn for themselves, a form of Artificial Intelligence called
``machine learning.'' Once robots are taught how to learn for
themselves, their decision-making process will be further removed from
human oversight.
What we can say, with certainty, is that bugs in commercial
software are frustratingly rampant, and take notoriously long for their
manufacturers to eradicate. Consumers are unlikely to tolerate becoming
``beta testers'' for driverless vehicles, serving as human guinea pigs
when the consequences are not a lost page of text but a loss of life.
2.2 Risk: Privacy and Security
Modern cars have become computers on wheels, collecting significant
amounts of data about the vehicle and the habits of the motorist that
drives them; some insurance companies have installed ``black boxes'' in
the vehicles they insure to track vehicle location, speed and other
metrics.\27\ By definition, evolving automation technologies will
collect, process and communicate vast amounts of information. The
recipients of the data stream will include, eventually, other vehicles
and likely the government agencies that operate the intelligent
transportation grid.
---------------------------------------------------------------------------
\27\ Markey, Sen. Edward J., ``Tracking and Hacking: Security and
Privacy Gaps Put American Drivers at Risk;'' https://
www.markey.senate.gov/imo/media/doc/2015-02-06_MarkeyReport-
Tracking_Hacking_CarSecurity%202.pdf, last accessed Nov. 28, 2015.
Consumer Watchdog has warned that ``black box'' data recorders in
vehicles could be misused in ways that threaten motorists' privacy. See
http://www.consumerwatchdog.org/feature/pay-you-drive-workshop-
comments-california-department-insurance.
---------------------------------------------------------------------------
The data is extremely valuable to hardware/software manufacturers
and insurance companies, but could prove costly for consumers.
Auto makers and software designers will want the data for
performance monitoring and safety improvement purposes, but
also to dispute their liability for crashes.
Google and other data collection companies will also want to
enhance the vast digital dossiers they already compile on each
American by including where motorists are going and what
they're doing, so advertisers can target their products, and
perhaps even subject motorists to continuous locality-based
advertising as their vehicle chauffeurs them through the
streets.
Insurance companies will seek data from cars to determine
who was at fault in an accident. But, increasingly utilizing
``big data,'' insurance companies will also seek to use the
data they collect to make underwriting decisions, enabling them
to avoid certain customers they deem too risky--a form of the
notorious historical practice of redlining- and to set premiums
so as to maximize profits rather than price risks, a highly
controversial practice known as ``price optimization.'' \28\
---------------------------------------------------------------------------
\28\ Letter from Robert Hunter, Consumer Federation of America, to
state insurance regulators, December 9, 2016 (http:// consumerfed.org/
testimonial/cfa-calls-insurance-commissioners-attention-unfair-price-
optimization-practices/); ``CFA's Hunter Reacts in Actuarial Battle:
Allstate's Plan Is Price Optimization,'' Carrier Management, November
29, 2016 (http://www.carrier
management.com/features/2016/11/29/161466.htm).
As data becomes increasingly valuable, it increasingly becomes a
target. Data breaches involving the accounts of billions of users
reflect the hacking epidemic in recent years.\29\ These incidents
demonstrate that only to the extent the legal system imposes
significant financial liability for such breaches will data collectors
be motivated to undertake the expensive hardening of their systems to
prevent third-party data incursions.
---------------------------------------------------------------------------
\29\ K. Sheridan, ``Data Breaches Exposed 4.2 Billion Records In
2016,'' Information Week, January 25, 2017.
---------------------------------------------------------------------------
It's not just the data that is vulnerable in increasingly automated
vehicles. The interconnected vehicles of the future will themselves be
subject to criminal and even terrorist hijacking. In 2015, two security
researchers managed to remotely hack into a 2014 Jeep Cherokee from a
laptop ten miles away and disable critical functions such as the
accelerator, paralyzing the car.\30\ Fiat Chrysler had to recall 1.4
million vehicles to fix the Jeeps' vulnerabilities.\31\ Another
research firm reported its was able to remotely take control of a Tesla
Model S and unlock the door of the car, take over control of the
dashboard computer screen, move the seats, activate the windscreen
wipers, fold in the wing mirrors and apply the brakes while the vehicle
was in motion--from ten miles away. Tesla uploaded an over-air
``software update'' to fix the vulnerability--ten days later.\32\
---------------------------------------------------------------------------
\30\ U.S. Says Only Jeeps Had Hacker Vulnerability Via Radios,
Associated Press, Jan. 9, 2016 (http://www.wsj.com/articles/u-s-says-
only-jeeps-had-hacker-vulnerability-via-radios-1452372453).
\31\ ``Autonomous cars to have `thousands of security risks,' ''
Autocar, 12 September 2016 (http://www.autocar.co.uk/car-news/industry/
autonomous-cars-have-`thousands-security-risks').
\32\ S. Khandelwai, ``Hackers take Remote Control of Tesla's Brakes
and Door Locks from 12 Miles Away,'' Hacker News, September 20, 2016
(http://thehackernews.com/2016/09/hack-tesla-autopilot.html?utm_so. .
.e+Hackers+News+-+Security+Blog%29&_m=3n.009a.1326.en0ao
0609g.rxj.)
---------------------------------------------------------------------------
In March 2016, the F.B.I. issued a warning to vehicle manufacturers
stating: ``it is important that consumers and manufacturers are aware
of the possible threats and how an attacker may seek to remotely
exploit vulnerabilities in the future.'' \33\ The F.B.I. pointed out
that hackers could gain access through a vehicle's cellular, USB,
Bluetooth, or Wi-Fi Internet connections: ``An attacker making a
cellular connection to the vehicle's cellular carrier--from anywhere on
the carrier's nationwide network--could communicate with and perform
exploits on the vehicle via an Internet Protocol (IP) address.'' \34\
---------------------------------------------------------------------------
\33\ ``Motor Vehicles Increasingly Vulnerable To Remote Exploits,''
Federal Bureau of Investigation, March 17, 2016 (http://www.ic3.gov/
media/2016/160317.aspx#fn1).
\34\ Id.
---------------------------------------------------------------------------
A wide variety of criminal misconduct could be facilitated via
hacking of automated and fully autonomous vehicles, ranging from
smuggling, to kidnapping, to homicide. A systemic attack on the
intelligent highway system could result in catastrophic loss of life
and by orchestrating traffic jams, grind commerce literally to a halt,
with serious financial repercussions.
2.3 Risk: Regulatory Failure
Failure to properly regulate the safety of self-driving vehicles is
rapidly becoming another serious safety risk. Congress enacted the
National Traffic and Motor Vehicle Safety Act of 1966 (NTMVSA) fifty
years ago ``to reduce traffic accidents and deaths and injuries
resulting from traffic accidents.'' The analysis of the proposed
legislation by the U.S. Senate concluded that:
The promotion of motor vehicle safety through voluntary
standards has largely failed. The unconditional imposition of
mandatory standards at the earliest practicable date is the
only course commensurate with the highway death and injury
toll.\35\
---------------------------------------------------------------------------
\35\ Committee Report on S. 3005, The Traffic Safety Act of 1966,
June 23, 1966, at 274.
The central safety focus of the NTMVSA, and NHTSA's activities
since its creation, has been the promulgation, after careful study and
a public hearing process, of Federal Motor Vehicle Safety Standards.
However, in an unprecedented departure from its statutory mission,
NHTSA entered into an unprecedented ``voluntary agreement'' with twenty
auto manufacturers in March 2016 to allow the industry to self-regulate
the sale of three safety technologies, known as Automatic Emergency
Braking (AEB), that assist cars in braking to avoid or limit the damage
from collisions.\36\ NHTSA rejected a petition by Consumer Watchdog and
other consumer advocates to require manufacturers to install even
established safety technologies, such as Automatic Emergency Braking,
as standard equipment in light vehicles--which NHTSA itself has
acknowledged would prevent tens of thousands of deaths and serious
injuries annually.\37\
---------------------------------------------------------------------------
\36\ ``U.S. DOT And IIHS Announce Historic Commitment Of 20
Automakers To Make Automatic Emergency Braking Standard On New
Vehicles,'' NHTSA (Press Release) March 17, 2017(https://www.nhtsa.gov/
press-releases/us-dot-and-iihs-announce-historic-commitment-20-
automakers-make-automatic-emergency). See Consumer Watchdog's letter to
NHTSA criticizing this action: http://www.consumerwatchdog.org/
newsrelease/consumer-advocates-demand-federal-agency-act-auto-safety-
petition).
\37\ NHTSA, Request for Comments, New Car Assessment Program,
Docket No. NHTSA-2015-0119, p. 110-118. In 2016, Consumer Watchdog and
other advocates formally petitioned NHTSA to promulgate a safety
regulation making AEB standard equipment. Ultimately NHTSA denied the
petition. For more information: http://www.consumerwatchdog.org/search/
apachesolr
_search/NHTSA%20AEB?filters=type%3Anewsrelease).
---------------------------------------------------------------------------
The decision marked a radical departure from the agency's
traditional mission. NHTSA Administrator Mark Rosekind argued that,
``the agency cannot make vehicles safe simply by imposing new
regulations and handing down fines. . . . We're going to have to find
new tools--that means new collaborations, new partnerships.'' \38\
Referencing the voluntary agreement for the deployment of AEB
technology, NHTSA claimed that ``bypassing the regulatory process would
save three years in making automatic braking systems standard
equipment.'' \39\ ``The unprecedented commitment means that this
important safety technology will be available to more consumers more
quickly than would be possible through the regulatory process.'' \40\
---------------------------------------------------------------------------
\38\ Exclusive: U.S., Major Automakers to Announce Safety Accord
Friday, David Shepardson, Detroit News, January 11, 2016).
\39\ J. Peltz, ``Automakers agree to make automatic braking a
standard feature by 2022,'' Los Angeles Times March 17, 2016 (http://
www.latimes.com/business/autos/la-fi-hy-auto-safety-20160317-
story.html).
\40\ Id.
---------------------------------------------------------------------------
As recently as 2013, NHTSA had adopted a go-slow approach to self-
driving vehicles, concluding, ``At this point, it is too soon to reach
conclusions about the feasibility of producing a vehicle that can
safely operate in a fully automated (or `driverless') mode in all
driving environments and traffic scenarios.'' \41\
---------------------------------------------------------------------------
\41\ NHTSA, Preliminary Statement of Policy Concerning Automated
Vehicles, May 15, 2013, p. 3.
---------------------------------------------------------------------------
But in early 2016, the Secretary of the Department of
Transportation told reporters that he ``wants to ease some of the
regulatory restraints to make it easier for the technology to
develop.'' \42\
---------------------------------------------------------------------------
\42\ POLITICO, Pro Transportation Report, Friday January 15, 2016.
---------------------------------------------------------------------------
In September 2016, NHTSA issued a 116 page ``Federal Automated
Vehicle Policy,'' which called upon manufacturers of automated and
self-driving vehicles to ``self-certify'' that they have considered a
fifteen point ``checklist'' of issues related to driverless
vehicles.\43\ NHTSA's Guidance leaves the evolution and deployment of
automated vehicles to hardware and software manufacturers, where it
will remain shrouded in secrecy and outside the purview of the public
generally, and motorists in particular.
---------------------------------------------------------------------------
\43\ NHTSA, Federal Automated Vehicles Policy, September 2016, p.
15-16.
---------------------------------------------------------------------------
If NHTSA's abdication of its safety responsibilities continues, the
introduction and deployment of autonomous technologies will proceed on
a manufacturer-by-manufacturer basis, without any enforceable,
industry-wide standards. Without industry-wide standards, the cost of
safety features will be prohibitive for all but the wealthiest
consumers. The deregulated deployment of automated vehicles will
exacerbate safety, liability and insurance issues.
2.4 Current Status of Robot Cars
No fully autonomous passenger vehicle is presently approved for
sale, much less being marketed. While proponents insist robot cars are
right around the corner, more objective observers expect a step-by-step
progression toward greater automation of vehicle functions--but with
the driver required ultimately to assume control.
Google, whose robot car unit is now known as Waymo, began testing
self-driving cars in 2009. California enacted AB 1298 in 2012 requiring
the Department of Motor Vehicles to enact self-driving vehicle
regulations. Rules covering testing robot cars took effect in 2014
requiring a test driver behind a steering wheel. The regulations
required companies to get a permit, report any crashes within 10 days
and file annual disengagement reports explaining when the self-driving
technology being tested failed. Most of what we know about testing
activities in California comes from the reports which the DMV, after
pressure from Consumer Watchdog and others, posts on its website.
Currently 21 companies have permits to test robot cars in California.
The firms have refused requests to disclose other important
information, including on board video and telemetry, from their
testing.
The disengagement reports demonstrate that the self-driving
vehicles are not ready to be deployed, at least without human drivers
behind a steering wheel who can seize control when the self-driving
technology fails. The latest report shows that Google/Waymo's test
vehicles logged 635,868 miles and the human test driver had to
intercede 124 times.\44\ In the past, the company has said that its
robot cars had difficulty correctly perceiving commonplace ``threats''
such as potholes, rain, wind and overhanging tree branches.\45\ There
were also software glitches and instances when the human test driver
took over because the robot car made an unwanted maneuver.\46\ In the
February 2016 sideswipe of a city bus by a Google robot test car in
northern California, requiring the bus to stop and its passengers to
disembark, Google claimed that the accident was a ``misunderstanding''
and a ``learning experience.'' \47\ Delphi's 2016 report stated that
its two test robot cars drove 3,125 miles in self-driving mode and had
experienced 178 ``disengagements.'' Reasons given for disengaging
included: construction zones; problems changing lanes in heavy traffic;
poor lane markings; the presence of emergency vehicles, pedestrians,
cyclists; failure to detect a traffic light and unexpected behavior
from another driver.\48\ Mercedes, which has asserted it will deploy an
autonomous vehicle by 2020, reported 336 disengagements in 673
miles.\49\
---------------------------------------------------------------------------
\44\ ``California Robot Car Disengagement Reports Show Technology
Not Ready for Safe Deployment Without Human Driver Behind Steering
Wheel to Take Control, Consumer Watchdog Says,'' February 1, 2017
(http://www.consumerwatchdog.org/newsrelease/california-robot-car-
disengagement-reports-show-technology-not-ready-safe-deployment-wit).
\45\ ``Consumer Watchdog Cites Shortcomings In Driverless Car
Technology; Says DMV Rules For Robot Cars Must Require Steering Wheel
So Human Drivers Can Take Over,'' March 19, 2015 (http://
www.consumerwatchdog.org/newsrelease/ consumer-watchdog-cites-
shortcomings-driverless-car-technology-says-dmv-rules-robot-cars).
\46\ See footnote 44.
\47\ M. McFarland, ``For The First Time, Google's Self-Driving Car
Takes Some Blame For A Crash,'' Washington Post, February 29, 2016
(https://www.washingtonpost.com/news/innovations/wp/2016/02/29/for-the-
first-time-googles-self-driving-car-takes-some-blame-for-a-crash/
?utm_term=.f19f6bdc6f4d).
\48\ See footnote 44.
\49\ Id.
---------------------------------------------------------------------------
Tesla reported two fatal crashes in 2016 (one in Florida, one in
China), both of which occurred while Tesla's ``Auto Pilot'' feature,
which the company famously introduced in 2014, was engaged.\50\ Tesla
continues to deceptively refer to its automated system ``Auto Pilot,''
though after the fatalities, it reprogrammed its Auto Pilot software
to, among other changes, disengage unless the driver touches the
steering wheel at regular intervals, indicating they are monitoring the
vehicle.\51\ Surprisingly, NHTSA later cleared Tesla of responsibility
for the Florida fatality, but an agency spokesperson incongruously
noted that humans must still manually pilot a Tesla equipped with Auto
Pilot: ``Autopilot requires full driver engagement at all times.'' \52\
---------------------------------------------------------------------------
\50\ N. Boudette, ``Self-Driving Tesla Was Involved in Fatal Crash,
U.S. Says,'' New York Times, June 30, 2016 (https://www.nytimes.com/
2016/07/01/business/self-driving-tesla-fatal-crash-investigation.html);
N. Boudette, ``Autopilot Cited in Death of Chinese Tesla Driver,'' New
York Times, September 14, 2016 (https://www.nytimes.com/2016/09/15/
business/fatal-tesla-crash-in-china-involved-autopilot-government-tv-
says.html?_r=0)
\51\ F. Lambert, ``Tesla Autopilot `Nags' And Restrictions Under
V8.0 Software Update--Breakdown,'' Electrek, September 22, 2016
(https://electrek.co/2016/09/22/tesla-autopilot-nags-and-restrictions-
under-v8-0-software-update-breakdown/).
\52\ N. Boudette, ``Tesla's Self-Driving System Cleared in Deadly
Crash,'' New York Times, January 19, 2017 (https://www.nytimes.com/
2017/01/19/business/tesla-model-s-autopilot-fatal-crash.html). See
Consumer Watchdog's criticism of this decision: http://www.consumer
watchdog.org/newsrelease/``midnight-action''-nhtsa-ends-probe-fatal-
tesla-florida-crash-accepting-company's-propaga.
---------------------------------------------------------------------------
In the rush to market of so-far unregulated robot technologies, the
adequacy of the current testing paradigm is questionable. Google/Waymo
claims that its computer-controlled vehicles have logged 300 years of
human driving experience. But the testing that would be required in
order to match the safety tolerance of commercial airplanes is
estimated at over one hundred millennia.\53\ A lower level of safety--
``a level of 80 percent confidence that the robotic vehicle is 90
percent safer than human drivers on the road,'' would still require 11
billion miles of testing (or about 5,000 years), according to
researchers at the University of Michigan, which is why they are
looking to shortcut the testing process, at least partly through
computer simulations.\54\
---------------------------------------------------------------------------
\53\ See Smith, Automated Driving & Product Liability, footnote 22,
p. 31.
\54\ S. Collie, ``Focusing On Tricky Situations Could Accelerate
Testing Of Self-Driving Cars,'' New Atlas, May 29, 2017 (http://
newatlas.com/university-michigan-self-driving-testing-acceleration/
49768/).
---------------------------------------------------------------------------
2.5 Current Status of Intelligent Transportation Infrastructure
The system of vehicle-to-vehicle, vehicle -to-satellite, vehicle-
to-road sensor communications--collectively referred to as vehicle-to-
everything, or ``v2e''--infrastructure that would permit tens of
millions of vehicles to simultaneously and securely operate without
human intervention is not even in the planning stage.
Studies of the technology are underway, but NHTSA only just
proposed uniform standards needed to ensure that all vehicles can
connect with each other regardless of manufacturer in December 2016,
and the proposal faces opposition from telecommunications companies
that want to use wireless channels for other purposes.\55\ How to
include pedestrians in such a system has not been resolved. Nor is
there any consensus on how to construct such a system--much less how
local, state and the Federal Government will cover the cost of
upgrading the 4.12 million miles of roadway in the United States.
---------------------------------------------------------------------------
\55\ R. Beene, ``Federal V2V Mandate Meets Growing Resistance,''
April 17, 2017 (http://www.autonews.com/article/20170417/OEM06/
170419865/federal-v2v-mandate-meets-growing-resistance).
---------------------------------------------------------------------------
2.6 The ``Driverless Divide''
The affordability of automated vehicles (and the cost of insuring
them) is an important safety issue in its own right, with profound
consequences when it comes to assessing the impact of autonomous
vehicles on liability and insurance.
Because no autonomous passenger vehicles are presently for sale,
any discussion of pricing is speculative. However, the price of robot
cars will directly affect the rate of deployment of the vehicles; the
higher the price, the fewer the number of people who will be able to
afford them. Those who cannot afford them will continue to operate
traditional cars that lack at least some safety features, placing them
at some correspondingly greater risk in the event of a crash.
Deployment will be further reduced because of NHTSA's abdication of
its regulatory responsibilities, discussed above. This is because the
Nation's auto safety regulator indicated through its 2016 ``Federal
Automated Vehicle Policy'' that it intended to rely on industry self-
regulation for self-driving vehicles, rather than promulgate formal
Federal Motor Vehicle Safety Standards (FMVSS) that would require all
new vehicles be equipped with the fully autonomous capability as
standard equipment.\56\
---------------------------------------------------------------------------
\56\ NHTSA, Federal Automated Vehicles Policy, September 2016.
---------------------------------------------------------------------------
Mandatory Federal safety standards create manufacturing economies
of scale from mass production that dramatically reduce the price of the
technology. Automakers resist industry-wide safety standards because
they can then treat expensive safety innovations as options to be
introduced in their most expensive vehicles, for which such options are
priced at a premium. It is not until the features become mandated
through the FMVSS process that they are rolled out in all vehicles
fleet-wide, and manufacturers drop the price. Thus the cost of cars
equipped with higher levels of automation will likely put them out of
reach of all but the wealthiest motorists.
Other price factors that will affect broad deployment will be car
repair and insurance premiums. Present day automotive electronics,
though increasingly complex, are relatively simple compared to the
technologies that will be needed to even partly automate passenger
vehicles. However, they have significantly raised the cost of repairs
(and insurance) for cars of more recent vintage. \57\
---------------------------------------------------------------------------
\57\ See, e.g., N. Brown, ``Conventional Car Repair Costs
Increased, While Hybrids Saw a Decrease,'' CleanTechnica.com, April 13,
2013 (http://cleantechnica.com/2013/04/15/conventional-car-repair-
costs-increased-while-hybrids-saw-a-decrease/); S. Finlay, ``More
Technology in Cars Increases Repair Costs,'' WardsAuto.com, September
17, 2013 (http://wardsauto.com/ dealerships/more-technology-cars-
increases-repair-costs); J. Selingo, ``Repair Shops See Roadblocks Set
by Dealers,'' New York Times, October 25, 2006 (http://www.nytimes.com/
2006/10/25/automobiles/autospecial/25repair.html ).
---------------------------------------------------------------------------
In other words, at least for the foreseeable future, there will be
the equivalent of what, in the early era of personal computing, was
described as a ``digital divide'': a significant disparity among
Americans between those who can afford vehicles with substantial
automation capabilities and those who cannot.
2.7 The ``Hybrid Highway''
There are diverging estimates of the date when a fully autonomous
vehicle--one that requires no human intervention--will be marketed to
the American public. However, any objective analysis demonstrates that
America is decades away from a transportation system that is completely
automated: one in which all vehicles on the road operate autonomously,
and there are no human drivers, no steering wheels, no brakes, nor
other human-based control devices; in which cars are in constant
electronic communication with each other, with ``intelligent'' road
systems built and maintained by municipal, state and Federal
governments, and with pedestrians equipped with their own electronic
devices.
The average age of vehicles on the road today is estimated at 11.5
years.\58\ Thus, even if a vehicle capable of operating under all
conditions without any human involvement (and absent the assistance of
intelligent highway infrastructure) were to come to market far sooner,
such vehicles will constitute a very modest percentage of the total
number of vehicles on the road.
---------------------------------------------------------------------------
\58\ Hirsch, Jerry, ``Better Quality Raises Average Age Of Cars On
U.S. Roads To 11.5 Years,'' Los Angeles Times, http://www.latimes.com/
business/autos/la-fi-hy-ihs-average-car-age-2015
0729-story.html last accessed Nov. 4, 2015.
---------------------------------------------------------------------------
Thus, it is clear that there will be a lengthy period in which
motorists and robots will share the roads in a hybrid system of human-
driven and highly automated, if not autonomous, vehicles.\59\
---------------------------------------------------------------------------
\59\ The Insurance Institute for Highway Safety and the Highway
Loss Data Institute concurs: ``Vehicles with humans at the wheel still
will dominate the fleet for many years. `Even if the U.S. Government
were to require all new vehicles sold to be autonomous tomorrow, it
would take at least 25 years until nearly 95 percent of the vehicles on
the road would have the capability.' '' ``Robot cars won't retire
crash-test dummies anytime soon,'' Status Report, Vol. 51, No. 8,
November 10, 2016 (http://www.iihs.org/iihs/news/desktopnews/driver-
seat-robot-cars-wont-retire-crash-test-dummies-anytime-soon).
---------------------------------------------------------------------------
This ``hybrid highway'' period will feature complex, potentially
dangerous interactions between people (motorists and pedestrians),
computer-driven cars, trucks and buses, remote-controlled drone
vehicles, and eventually the so-called ``intelligent'' public streets
and freeways that are supposed to help them all navigate safely.
3.0 The Personal Responsibility System and Self-Driving Vehicles
3.1 Tort liability
The judicial branch is responsible for interpreting and applying
laws. However, state courts \60\ also play a unique legislative role:
they are the source of what is known as ``common law.'' Originating
from ancient English law, and often dating back to the formation of the
United States, common law is a body of case decisions issued by state
courts that defines rights and remedies in the absence of any
underlying statutory authority. State legislatures have the authority
to amend or even repeal the state's ``common law,'' and they frequently
do so.
---------------------------------------------------------------------------
\60\ The sole exception is Louisiana, which is known as a ``civil
law'' state. In Louisiana, courts lack any authority to adjudicate a
matter absent a statute.
---------------------------------------------------------------------------
A tort is a wrongful act that causes bodily injury or property
damage. The common law of torts is a collection of legal rights,
responsibilities and remedies developed and applied by civil courts
when a wrongful act has caused harm. The purpose of tort law is to
expose wrongdoing, compensate victims of the wrongdoing, punish the
wrongdoers and deter future wrongdoing.
3.1.1 Negligence
Generally, tort liability is predicated upon the following judicial
determinations: (1) the defendant owed to the plaintiff a duty of
reasonable care; (2) the defendant breached that duty (3) the breach
caused damage to the plaintiff. Under the Personal Responsibility
System established by tort law, a person or company who committed a
tort is liable for the injuries, property damage, lost wages, physical
pain, emotional damage any and other kind losses that arise as a
result. Intentional wrongdoing that is considered particularly
egregious or oppressive may be punished by punitive damages: the
wrongdoer is penalized for such misconduct.
Disputes over torts are typically adjudicated through the civil
court system, which is the practical embodiment of the common law right
to a trial by jury, one of America's most hallowed traditions. However,
as discussed below, auto insurance has evolved as a mechanism for
ensuring compensation without necessity of bringing legal action in
modest disputes.
3.1.2 Product liability
A separate set of consumer friendly rules has evolved for
relatively more rare torts involving products that are considered
``inherently'' dangerous, such as cars. In California, for example, a
defendant is held strictly liable for injuries caused by such products,
when a product was used in intended or reasonably foreseeable manner
(includes reasonably foreseeable misuse, abuse, changes, alterations,
etc.); was in defective condition when it left defendant's possession;
and the defective product was the legal cause of the plaintiff's
injuries or damages.
Unlike regular negligence cases, in product liability disputes the
injured consumer is not required to prove that the defendant was
negligent, i.e., that the defendant failed to exercise reasonable care,
or intended to cause harm. The public policy behind this variation in
tort law is that it would be prohibitively difficult and expensive for
a consumer to prove that the manufacturer of a product was careless in
making the product, nor would a consumer have the ability to determine
whether the product was defective prior to purchase. The protections of
strict liability rules have been extended to include entities that re-
sell or distribute the products. However, defects in road design,
construction and maintenance are sometimes governed by more restrictive
state statutes.
It is widely assumed that as vehicle automation progresses, and
motorists cede driving functions to the vehicle's computer systems,
responsibility will shift from motorists to manufacturers of the
hardware and software, and claims will be adjudicated under product
liability law.
3.1.3 Common Carrier Liability
Another long established common law principle is common carrier
liability. Common carriers are companies that transport people (or
goods) pursuant to a license provided by a government agency. Common
Carriers include taxis, buses and ferries. Common Carriers are held to
a very high legal standard. Under California law, for example, ``A
carrier of persons for reward must use the utmost care and diligence
for their safe carriage, must provide everything necessary for that
purpose, and must exercise to that end a reasonable degree of skill.''
``Common carriers are responsible for any, even slightest, negligence
to passengers and are required to do all that human care, vigilance,
and foresight reasonably can do under all the circumstances.''
3.1.4 Liability Scenarios
As noted above, human-operated vehicles will remain the predominant
form of personal transportation for the foreseeable future. Vehicles
with wide disparities in the level of onboard technology will share the
roads with newer vehicles containing an equally wide variety of the
more sophisticated automation technologies. The intelligent
infrastructure of vehicle, satellite and road communications that many
view as integral to the safety of an autonomous transportation system
has yet to be planned, much less constructed, and will not play any
significant role for the foreseeable future.
This Hybrid Highway will be the product of a hugely complex system
of hardware and software built, marketed, maintained and operated by
corporations manufacturing hardware and software, engineers, software
programmers, public agencies as well as motorists.
Compounding the threat matrix are vehicle security failures,
ranging in consequence from privacy breaches to criminal or terrorist
hacking; the absence of Federal safety rules to standardize
technologies; wealth based disparities in the affordability of
autonomous technology. Flaws and failures in any single aspect of this
complex environment could lead to death, injury and property damage.
Even in a distant theoretical future in which all vehicles are
controlled by robots, the same concerns apply.
The table below is based on the taxonomy for self-driving vehicles
published by the Society of Automotive Engineers (SAE),\61\ which has
been broadly endorsed as a tool for discussion of these issues. The
table illustrates who will be responsible under some likely risk and
liability scenarios; the leftmost column describes the SAE level of
automation and the top row lists particular liability risks.
---------------------------------------------------------------------------
\61\ ``Surface Vehicle Recommended Practice,'' Society of Vehicle
Engineers (J3016), September 2016.
As the table illustrates, there is no scenario in which disputes
will not require resolution through the civil justice system. Note the
SAE taxonomy explicitly assumes that the vehicle may issue a ``request
to intervene'' to a human occupant of a fully self driving vehicle
(though in the highest automation modes, the robot car or truck will
not have an ``expectation that a user will respond'' to such a
request). It is implicit in this analytical framework that the vehicle
will contain the necessary equipment (steering wheel, brake pedal,
etc.) that will enable the occupant to seize control. In other words,
the SAE framework envisions no scenario in which a human cannot
ultimately obtain control of a robot car. No company planning to sell
robot vehicles has stated whether they will come equipped with the
complement of control devices present in traditional vehicles today.
(For purposes of shifting liability to consumers, manufacturers of self
driving vehicles may choose to retain those devices.)
Assume, for example, that a vehicle is capable of operating
autonomously, but a passenger is still expected to seize control of the
vehicle in some circumstances (Level 3-4 under the SAE taxonomy). That
person, presumably clearly designated as such by the vehicle itself,
will remain subject to liability for failure to intercede properly.
Assume a vehicle is capable of full autonomous operation, but a
person in the vehicle is still able to request that the vehicle
``surrender control'' (SAE Level 5). Or assume that the vehicle issues
a request for the user to intervene--even though the user is not
expected to. In the event of a crash, a person's failure to demand
control, or agree to accept control, could itself be the basis for
liability.
And as automated technologies become more sophisticated, and cars
and trucks are able to operate autonomously from human intervention,
manufacturers of the hardware and software will face strict liability
for design or manufacturing defects that caused a crash.
In all of five of the SAE scenarios, third parties, including
manufacturers, will be permitted to dispute whether the vehicle, or the
motorist, was responsible. Facing strict liability for crashes,
manufacturers will certainly have an incentive to dispute their
responsibility. And it's worth noting that in several of the most
highly publicized crashes involving Tesla to date, the company has been
reluctant to accept full responsibility. In a highly-publicized
accident in which a Tesla owner died when his Tesla, on Autopilot,
failed to recognize a truck crossing the road, Tesla went so far as to
release ``black box'' data from a vehicle to support Tesla's position
that the driver was at fault, not the car.\62\ Similarly, the
ridesharing firm Uber blamed test drivers when its vehicles, illegally
operating in self-driving mode, were caught running red lights in San
Francisco.\63\
---------------------------------------------------------------------------
\62\ See, for example, D. Shephardson, ``Google Takes the Blame--
Sort of--for Its Self-Driving Car Crash,'' Reuters, February 29, 2016
http://time.com/money/4242030/google-self-driving-car-crash-fault/);
``Tesla Says Autopilot Not to Blame in Crash With Bus in Germany,''
Reuters, September 29, 2016 (http://fortune.com/2016/09/29/tesla-
autopilot-crash-germany/); C. Davies, ``In New Model X Crash, Tesla
Suggests Autopilot Not To Blame,'' Slash Gear, July 6, 2016 (https://
www.slashgear.com/in-new-model-x-crash-tesla-suggests-autopilot-not-to-
blame-0644
7354/).
\63\ ``Witness Saw Uber Robot Car Drive Through Red Light Three
Weeks Ago in San Francisco,'' Consumer Watchdog, December 20, 2016
(http://www.consumerwatchdog.org/newsrelease/witness-saw-uber-robot-
car-drive-through-red-light-three-weeks-ago-san-francisco); J. Hood,
``California Orders Uber's Self-Driving Cars Off The Road,'' Consumer
Affairs, December 15, 2016 (https://www.consumeraffairs.com/news/
california-orders-ubers-self-driving-cars-off-the-road-121516.html).
---------------------------------------------------------------------------
While a self-driving vehicle will collect vast amounts of data that
will potentially offer enormous insight into the reasons for a crash,
key questions may not be answered by that data. For example, did the
vehicle correctly inform the designated passenger of the status, such
that the passenger should have known to assume control? Did the vehicle
fail to request human intervention? Did the vehicle improperly reject a
user's demand for surrender of control? Would it matter to the inquiry
what the passengers were doing at the time of the crash? It is not
clear to what extent the vehicle will collect all the data necessary to
determine what happened in the seconds before a crash. Will there be a
continuously recording camera and microphone in the passenger
compartment, such that third parties could argue the passenger was
distracted?
Consumers who own or lease self-driving vehicles will face far
greater maintenance responsibilities than they bear today. For example,
self-driving vehicles will be reliant on external sensors--the eyes and
ears of the robot car's brain. A scrape or dent that impairs a sensor
while the vehicle is in the driveway could lead to deadly consequences
when the vehicle is on the road.
Moreover, the computer brains of robot cars will inevitably require
software updates. What happens if the consumer fails to download a
software update, or visit the dealership if that is required? No
potential manufacturer of a self driving vehicle has offered to assume
liability for a crash caused, even partly, by the consumer's failure to
maintain the vehicle's operational status.
Each of these scenarios confirms that an inquiry into a consumer's
``fault'' will be necessary even in the era of fully autonomous
vehicles.
Complicating these scenarios is the fact that hardware and software
manufacturers consider their technology proprietary; indeed, for
security reasons, it may be impossible for even the owner to access any
vehicle data.
Finally, the manufacturers of automated vehicles acknowledge their
self-interest when it comes to liability. Not one manufacturer has
agreed to assume all liability for the harm caused by their automated
vehicles. Three companies have been quoted as stating that they will
accept legal liability when their cars are in fully autonomous mode:
Volvo,\64\ Mercedes and Google.\65\ But news reports indicate that
Mercedes and Google added a salient limitation on their pledge: that
``their technology is at fault.'' \66\ Of course, that caveat will
leave the owner of the robot car exposed to liability in cases where
the manufacturer insists the crash was not the fault of its hardware or
software--necessitating an inquiry into the drivers' fault.
---------------------------------------------------------------------------
\64\ M. Harris, ``Why You Shouldn't Worry About Liability for Self-
Driving Car Accidents,'' IEE Spectrum, October 12, 2015 (http://
spectrum.ieee.org/cars-that-think/transportation/self-driving/why-you-
shouldnt-worry-about-liability-for-selfdriving-car-accidents).
\65\ M. Ballaban, ``Mercedes, Google, Volvo To Accept Liability
When Their Autonomous Cars Screw Up,'' Jalopnik, October 7, 2015
(http://jalopnik.com/mercedes-google-volvo-to-accept-liability-when-
their-1735170893).
\66\ Id.
---------------------------------------------------------------------------
The search for truth and justice in such circumstances will require
the full powers of the civil justice system. The right to challenge
corporate mistakes and reckless profit-driven conduct, in an impartial
judicial forum with all the procedural protections of the civil justice
system, starting with trial by jury, and including the strict liability
of hardware and software manufacturers, will be critical.
3.2 Insurance
The determination of fault and compensation for injury and property
damage are matters made by courts. However, the evolution of the
automobile as the predominant form of transportation in the United
States led to the establishment of mandatory minimum auto insurance
coverage requirements--known as ``compulsory financial responsibility
laws''--in Massachusetts in 1927; today every state but New Hampshire
requires such coverage. Thus motorists, as a condition of owning or
leasing a vehicle for operation on public roads, must buy insurance
that will cover, to at least a minimum extent, that motorist's
liability should he or she cause injury or damage to another person or
their property. California, for example, requires most motorists to
obtain a policy that would pay up to $15,000 in bodily injury
compensation per person (for a maximum of $30,000 among all injured
parties) and $5,000 in property damage.\67\
---------------------------------------------------------------------------
\67\ Other insurance coverage, though typically optional, is often
purchased by consumers to protect their own vehicles against fire or
weather damage (comprehensive coverage), or crashes that don't involve
a third party--such as with a tree or other object (known as collision
coverage). In states where many motorists operate without insurance,
consumers often find it prudent to purchase ``uninsured motorist''
coverage, so that if they are hit by an uninsured motorist, their
expenses are covered.
---------------------------------------------------------------------------
In the event of a crash, persons who suffer loss or damage as a
result of the at-fault driver make a claim upon the at fault driver's
insurance coverage. The insurance company is required to make an
objective determination of the fault of its insured (the exact
requirements for that determination vary depending upon state law, and
in California are governed by Proposition 103--see below), and pay the
claim.
Mandatory auto insurance coverage assures that motorists will have
the means to provide at least a minimum level of compensation for
modest accidents they cause--hence the term ``financial
responsibility.'' Absent such insurance, the at-fault motorist risks a
potentially devastating civil judgment against his or her home or other
assets. Auto insurance also alleviates what would otherwise be a
significant burden on courts to adjudicate even minor disputes
involving car accidents.
The cost of insurance and the underwriting and marketing practices
of insurance companies have long been a source of public
dissatisfaction and are often highly controversial. Regulation of
insurance rates and practices is a matter of state law. The requirement
that motorists purchase third party insurance coverage from private
insurance companies has necessitated the establishment of consumer
protections to assure that consumers are treated in a fair and non-
discriminatory fashion when buying insurance, and in the event an
insurance claim has been filed. However, the degree of protections
afforded consumers varies sharply from state to state, as a 2013 report
by the Consumer Federation of America found.\68\
---------------------------------------------------------------------------
\68\ ``What Works? A Review of Auto Insurance Rate Regulation in
America,'' Consumer Federation of America, November 12, 2013.
---------------------------------------------------------------------------
3.2.1 Insurance and autonomous vehicles
In 2015, United States-based insurance companies held a total of
$8.4 trillion in assets.\69\ They wrote roughly $192 billion net auto
insurance premiums nationwide in 2015 (not including commercial
insurance) and projected they would pay $145 billion in claims.\70\
---------------------------------------------------------------------------
\69\ ``Annual Report On The Insurance Industry,'' Federal Insurance
Office, U.S. Department Of The Treasury (September 2016), p. 12.
\70\ Insurance Information Institute (http://www.iii.org/fact-
statistic/auto-insurance, last visited June 12, 2017).
---------------------------------------------------------------------------
The insurance industry initially appeared to view self-driving
vehicles as an existential threat. Within the insurance industry, there
has been frequent speculation, sometimes verging on panic, at the
prospect of that revenue stream evaporating with the advent of
accident-free, driverless vehicles: if there are no accidents, the
industry reasoned, then why would anyone buy insurance?
With the benefit of several years of hindsight, the insurance
industry's immediate fears appear to have subsided. Under any
transportation system in which a consumer is or may be required to
operate a vehicle, or even simply to maintain it, state tort laws will
hold them accountable. Consumers will continue to purchase insurance
coverage to protect innocent third parties against injuries or property
damage and to cover their own repair expenses.
Indeed, as automation technologies enable vehicles to operate
without human intervention, the makers of the vastly more complex
hardware and software will face increased tort liability for
defectively designed or manufactured products. These firms will seek to
purchase insurance product liability insurance coverage to pay such
claims. Self-driving cars and trucks will create new markets for
vehicle insurance coverage that do not exist today.
It is too early to know the full financial, economic or social
impacts of robot cars will be at this juncture. But we do know that
insurance coverage will remain an essential protection in the era of
driverless vehicles.
For consumers, the pricing of insurance, historically a significant
concern, is likely to become a major economic factor as vehicle
automation increases.
As noted previously, while it seems logical that the evolution of
auto safety systems will lead to fewer crashes, there is as yet no
evidence behind the surmise that robot cars will lead to an overall
reduction in crash frequency, severity or claims costs. The
incorporation of electronics in today's cars and trucks, though
rudimentary by comparison to the complex hardware and software needed
to maneuver vehicles without human drivers, have already spiked repair
costs and insurance premiums.\71\ The far greater cost of repairing
automated vehicles will likely lead insurance companies to dramatically
inflate the price of liability, collision and comprehensive insurance
coverage.
---------------------------------------------------------------------------
\71\ See Section 2.6.
---------------------------------------------------------------------------
Moreover, risks that today are not especially relevant to cars and
trucks--such as privacy, security or even mass terrorism--will be much
more of a threat to robot vehicles. Insurance companies will likely
assess the heightened risk/threat matrix of the new and untested
technologies and the hybrid highway as a basis to argue for substantial
rate increases in the near term.
Finally, there is a very real danger that insurance companies will
pursue a new form of ``redlining'' to favor motorists who can afford
more expensive cars with expensive computer-based systems and
discriminate against those who cannot by refusing to sell them
insurance, or adding surcharges to the price of insurance -practices
with pervasive historical antecedents in the insurance industry.
Strengthened consumer protections against excessive insurance
premiums will prove crucial for as insurance companies price the risk
of highly automated vehicles--particularly since state insurance
regulators often lack the authority (or desire) to bar abusive rates
and practices.
3.2.2 The Proposition 103 Model
According to a 2013 report by the Consumer Federation of America,
``California stands out from all other states in having the best
regulatory system for protecting consumers.'' \72\ Enacted by
California voters in 1988, California's insurance reform law provides
precisely the stronger protections consumers will require in the era of
robot vehicles.
---------------------------------------------------------------------------
\72\ In 1984, the California Legislature amended its financial
responsibility law to address the growing number of uninsured
motorists. The amendment allowed police officers to request proof of
insurance and to cite those who did not produce it. While Californians
were required by law to purchase insurance, California's insurance law
did not require insurance companies to sell it to all individuals; nor
were there any limits on the price insurance companies charged. Many
Californians could not afford to purchase auto insurance, particularly
in neighborhoods that were subject to insurance ``redlining,'' even if
it was available. The inequities of the mandatory purchase requirement,
combined with escalating auto, home and business insurance premiums,
sparked a voter revolt that led to the passage of Proposition 103 in
November, 1988. The measure (Insurance Code section 1861.1 et seq.)
fundamentally rewrote California's insurance laws. For a detailed
discussion of the origin, purposes and provisions of Proposition 103,
see Harvey Rosenfield, Auto Insurance: Crisis and Reform, 29 University
of Memphis Law Review 69 (Fall 1998). Much more information about
Proposition 103 is available at www.Consumer
Watchdog.org.
Review of insurance rates. Proposition 103 applies to
automobile, homeowner, business, and all other property-
casualty insurance. It mandated a one-time rollback to November
1987 levels and a further 20 percent reduction in premiums.
Over $2 billion in refunds were paid by insurance companies
under this directive. The measure requires all property-
casualty insurance companies to open their books and justify
existing or proposed rate changes, subject to stringent
controls on insurance company profiteering, waste, and
inefficiency, and to obtain the Insurance Commissioner's
approval before such changes may take effect. Insurance
companies must show that their rates are based on verifiable
---------------------------------------------------------------------------
loss data and legitimate expenses.
Prohibition on anti-consumer and discriminatory practices.
The measure bars ``unfairly discriminatory'' rates or premiums.
It also subjects the insurance industry to lawsuits for
violation of Proposition 103's provisions and California's
civil rights, consumer protection and other laws.
Public disclosure and transparency. The law authorizes the
Insurance Commissioner to obtain any data--such as rate and
premium data--from insurance companies that is needed to
regulate their rates and practices. The Commissioner must
disclose to the public all information -that insurance
companies provide.
Public participation. The law authorizes and encourages
consumers to monitor and challenge existing rates, applications
for rate changes, or any other practices that may be unlawful,
either in the courts or before the California Department of
Insurance. Under certain conditions, the Insurance Commissioner
must hold a public hearing on such challenges. The law requires
insurance companies to pay the legal fees and expenses of
consumers who participate and make a ``substantial
contribution'' to the outcome of a legal proceeding. The law
also made the Insurance Commissioner, usually an appointed
position, an elected post.
Preventing insurance companies from seeking unjustified rate
increases will be critical as self-driving vehicles become more
commonplace, particularly because initially insurance companies will
have limited experience in assessing the risk they pose, and for that
reason alone will seek to inflate projections of future claims and the
cost of repairing or replacing vehicles.
Special protections against unfair automobile insurance
premiums. Particularly relevant to self-driving vehicles,
Proposition 103 established a special set of rules that govern
the pricing of automobile insurance.
Auto insurance premiums must be determined principally by three
specified rating factors--the insured's driving safety record;
annual mileage, and years of driving experience--and, to a
lesser extent, by any ``optional'' rating factors that ``the
commissioner may adopt by regulation and that have a
substantial relationship to the risk of loss.'' \73\ The use of
any other criterion constitutes unfair discrimination and is
unlawful.
---------------------------------------------------------------------------
\73\ See California Insurance Code section 1861.02(a). The current
list of authorized optional rating factors can be found at 10 CCR
2632.5(d).
Making the driver's own safety record the principal determinant
of premiums gives motorists a strong incentive to drive safely.
The measure further requires insurers to offer a 20 percent
good-driver discount to all qualifying consumers: individuals
with a virtually clean driving record (one moving violation is
permitted) for the preceding three years. This provides a
further incentive for careful driving.
Basing auto insurance premiums on a motorist's individual
responsibility, as reflected by their driving record, will
remain of paramount importance for consumers in the era of
self-driving vehicles, because in every conceivable scenario
the consumer may still bear potential liability in the event of
a crash.
As today, when a motorist is driving a vehicle, they bear
responsibility for any injuries or property damage for which
they are at fault. During times when the robot is driving the
vehicle, the consumer occupant will very likely still have a
legal duty to take control in the event of an imminent
accident. Even when a self-driving vehicle is parked, the
consumer will be responsible for maintaining it in proper
condition. A consumer's driving safety record will be based on
whether the automated car can avoid tickets and accidents in
all these circumstances. And, as noted, the hardware and
software manufacturers of automated vehicles will have a
financial motive to dispute fault. Because there will never be
a 100 percent guarantee that the occupant will not be
responsible for a traffic violation if a vehicle fails to
properly stop as a pedestrian enters a crosswalk or crosses
into an intersection in heavy traffic or if a vehicle's sensor
fails, or the computer is hacked, and a crash results, a
motorist's driving safety record should be the predominant
factor is setting premiums.
Similarly, annual mileage and years of driving experience,
along with several of the optional rating factors previously
adopted by the Commissioner, reflect the motorist's risk,
without regard to whether the policyholder is driving a car
equipped with automation technology. Cars equipped with
improved technology will be rated, as they are today under
Proposition 103, based on their repair or replacement cost for
purposes of comprehensive (weather damage, fire and theft) and
collision coverages.
Other optional rating factors that will remain applicable: the
percentage use of vehicle by rated driver; type of vehicle;
vehicle performance capabilities, including alterations made
subsequent to original manufacture; and vehicle
characteristics, including engine size, safety and protective
devices, the vehicle's vulnerability to damage, repairability,
and installed theft deterrent devices. Cars and trucks equipped
with improved technology will be rated, as they are today under
Proposition 103, based on their repair or replacement cost for
purposes of comprehensive (weather damage, fire and theft) and
collision coverages.
Assessing the overall impact of these reforms in its 2013 analysis,
the Consumer Federation of America determined that California was the
only state in the Nation where the average auto insurance premium went
down between 1989 and 2010, saving motorists alone over $100 billion in
premiums since the law took effect.\74\
---------------------------------------------------------------------------
\74\ ``What Works? A Review of Auto Insurance Rate Regulation in
America,'' Consumer Federation of America, November 12, 2013.
---------------------------------------------------------------------------
Apart from preventing price gouging and discriminatory practices,
Proposition 103 provides regulators and consumers with the tools and
methodology to address other issues raised by autonomous vehicles.
For example, as noted above, insurance companies collect
significant amounts of data about motorists; some have begun installing
black boxes in their vehicles to track mileage and other metrics.\75\
An even more troubling abuse is the recent phenomenon, previously
noted, of insurance companies utilizing the vast trove of personal data
collected by Google, Amazon, various credit bureaus and other firms to
individualize a motorist's premiums based on algorithms that consider
rating factors that have nothing to do with risk, such as the
likelihood that a particular consumer will accept a modest overcharge
without protest--a practice known as price optimization that is
unlawful in nineteen states and the District of Columbia.\76\ Under
Proposition 103, such practices can be challenged in court and
investigated by state Department of Insurance. Acting at the request of
the Los Angeles Superior Court, where a class action has been filed,
the California Department of Insurance is presently investigating
whether Farmers Insurance is engaged in the practice, which is unlawful
under Proposition 103.\77\
---------------------------------------------------------------------------
\75\ See footnote 27. In response to advocacy by Consumer Watchdog,
regulations promulgated pursuant to Proposition 103 bar insurance
companies from collecting data about the location of an insured
vehicle, except as part of an emergency road, theft, or map service.
See 10 CCR Sec. 2632.5(c)(2)(F)(i)5.
\76\ See Section 2.2.
\77\ In the Matter of the Rating Practices of Farmers Insurance
Exchange And Mid Century Insurance Company (File No. NC-2017-00003).
---------------------------------------------------------------------------
As another example, the evolution of the car industry into a more
frequent litigant may create conflicts in the duties the insurance
industry owes its policyholders. Some manufacturers of self driving
hardware and software may purchase large quantities of insurance
coverage against product liability suits. If so, significant conflicts
of interest may arise: if the same insurance company sells insurance
policies to motorists or owners of automated vehicles and to
manufacturers, the legal duty to handle its policyholders' claims in
good faith, which each insurance company owes its individual
policyholders, could well collide with its financial incentive to
protect the interests of the manufacturer that bought a product
liability policy.
In other words, in the era of self-driving vehicles, manufacturers
and insurance companies may have a vested financial interest in
protecting each other's bottom line, in which case the threat to
consumers when it comes to crashes is that every accident will be
treated as ``your fault.'' New rules to protect consumers against such
conflicts will likely be necessary. Proposition 103 provides the
Commissioner and the courts with the authority to adjudicate these
unexpected secondary effects in an open and transparent forum.
4.0 The Industry Agenda to Roll Back Consumer Rights
Over the last five decades, Americans have benefitted from a
paradigm change in consumer protection. Across the economy, rules have
been put in place to expand the rights of consumers exposed to
physically or financially injurious products or services. Many of these
laws, such as those barring and punishing false advertising, defective
products, sharp financial practices, have become deeply ingrained in
consumers' bedrock expectations of the marketplace.
These norms have long been the target of a national attack by
insurance companies, automakers and other powerful corporations, their
lobbyists, and sponsored allies in academia, seeking to restrict
consumer rights under the Personal Responsibility System. They are now
recycling discredited anti-consumer proposals to limit corporate
accountability, backed by big business, insurance companies and their
network of lobbyists and academics, that have failed throughout the
United States, and which California voters have rejected multiple times
at the ballot box (Propositions 101, 104 and 106 in 1988; Propositions
200, 201 and 202 in 1996).
As noted previously, the suggestion that the transformation to a
completely automated transportation system is imminent is a fantasy.
But it's a fantasy that automakers and insurance companies are now
attempting to exploit in order to press lawmakers to re-write consumer
protection laws in their favor.
To do so, they are replicating themes that have proven successful
in previous campaigns.
4.1 Restrictions on liability laws to encourage ``innovation''
Manufacturers of hardware and software are quietly proposing to
revise liability laws and rules so as to limit their financial
responsibility for deaths and injuries caused by their automated or
self-driving technology. Insurance companies, which profit primarily
through the investment of premiums, have a similar financial motive to
press for limits on liability, since the fewer and smaller claims
payouts leaves more premium dollars for insurance companies to invest,
particularly in states where regulators do not have the authority to
limit rate increases to reasonable projections of future losses.
Among the proposals advanced by manufacturers and insurance
companies are arbitrary caps on how much compensation juries can award
to victims of negligence or intentional misconduct that causes deaths
or injuries, and restrictions on how much attorneys can charge for
their representation of such victims.\78\
---------------------------------------------------------------------------
\78\ R. Nader, ``Suing for Justice,'' Harpers, April 2016.
---------------------------------------------------------------------------
A different approach, adopted by the George W. Bush Administration
although most certainly unconstitutional,\79\ called for NHTSA and
other Federal agencies to override state consumer protection laws.\80\
The Obama Administration later reversed it.\81\ (The Trump
Administration is reportedly preparing its own ``guidelines'' for self
driving vehicles; according to the new Secretary of the Department of
Transportation: ``We don't want rules that impede future technological
advances.'' \82\)
---------------------------------------------------------------------------
\79\ D. Vladek, ``The Emerging Threat of Regulatory Preemption,''
The American Constitution Society, January 2008.
\80\ M. Levin and A. Miller, ``Industries Get Quiet Protection From
Lawsuits,'' Los Angeles Times, February 19, 2006 (http://
articles.latimes.com/2006/feb/19/nation/na-preempt19)..
\81\ P. Rucker, ``Obama Curtails Bush Policy That Let Federal Rules
Override State Laws,'' Washington Post, May 22, 2009 (http://
www.washingtonpost.com/wp-dyn/content/article/2009/05/21/
AR2009052104016.html).
\82\ D. Shephardson, ``U.S. Plans To Update Self-Driving Guidelines
In Coming Months,'' Reuters, June 5, 2017 (http://www.reuters.com/
article/us-usa-selfdriving-idUSKBN18W2JR).
---------------------------------------------------------------------------
Often, arguments in support of such proposals are couched in a
threat: that absent such liability limits, manufacturers will not bring
a product to the American marketplace. Thus the liability protections
are described as ``impediments'' to innovation.
Perhaps the most illustrative example is the liability bailout of
the American nuclear power industry in the 1950s. After World War II,
Americans were enamored with atomic energy; the ``peaceful use'' of the
atom was heralded as providing electricity so inexpensive for American
households that it would be ``too cheap to meter.'' There was catch. In
what should have been understood as a grave warning sign of the risks
of nuclear power, the insurance industry claimed it could not provide
the insurance that the nascent atomic energy industry needed to cover
its potential liability for a nuclear meltdown or other accident.
Potential liability is what stood in the way of ``progress,''
supporters of nuclear power insisted. In 1957, Congress obligingly
passed the Price Anderson Act, which immunizes the atomic energy
industry from liability to the American public in exchange for a tiered
fund consisting of a contribution from the nuclear industry of up to
$13 billion, followed by an expected congressional bailout. Nuclear
power has proven to be an economic disaster for American taxpayers and
ratepayers.\83\ (The $13 billion limit on the nuclear industry's
liability is woefully inadequate: the Japanese government's latest
estimate of the cost of the 2011 meltdown at Japan's Fukushima Daiichi
Power Plant--still underway--is $188 billion.\84\)
---------------------------------------------------------------------------
\83\ R. Nader, ``Nuclear Power's Insanities--Taxpayer-Guaranteed,''
Common Dreams, September 06, 2014 (http://www.commondreams.org/views/
2014/09/06/nuclear-powers-insanities-taxpayer-guaranteed).
\84\ Y. Obayashi and K. Hamada, ``Japan nearly doubles Fukushima
disaster-related cost to $188 billion,'' Reuters, December 9, 2016
(http://www.reuters.com/article/us-tepco-fukushima-costs-
idUSKBN13Y047).
---------------------------------------------------------------------------
A report by the U.S. Chamber of Commerce's dedicated anti-liability
law unit insists that legal liability will ``chill this promising
technology [autonomous vehicles] and the huge advances in overall
public safety it promises.'' \85\ It continues: ``Where liability
exposure poses a threat to an emerging technology, legislators should
adopt reasonable constraints on liability.'' \86\
---------------------------------------------------------------------------
\85\ U.S. Chamber of Commerce Institute for Legal Reform, ``Torts
of the Future: Addressing the Liability and Regulatory Implications of
Emerging Technologies,'' March 2017, p.2.
\86\ Id., p. 54.
---------------------------------------------------------------------------
In a lengthy paper on legal liability and self driving vehicles
published in 2016, RAND Corporation, which has received substantial
funding from the insurance industry and has been a long-time advocate
of restrictions on victim compensation rules, makes the same point:
Current liability laws may well lead to inefficient delays in
manufacturers introducing AV [autonomous vehicle] technologies.
The gradual shift in responsibility for automobile operation
from the driver to the vehicle may lead to a similar shift in
liability for crashes from the driver to the manufacturer.
Recognizing this effect, manufacturers may be reluctant to
introduce technology that will increase their liability.\87\
---------------------------------------------------------------------------
\87\ J. Anderson, et al., Autonomous Vehicle Technology: A Guide
for Policymakers, RAND, 2016, p. 118.
---------------------------------------------------------------------------
4.2 Restrictions on liability laws to ``lower insurance rates''
The insurance industry is now resurrecting a long-abandoned and
discredited scheme known as ``no fault'' auto insurance. RAND's report
concludes: ``Th[e] shift in responsibility from the driver to the
manufacturer may make no-fault automobile-insurance regimes more
attractive.'' \88\
---------------------------------------------------------------------------
\88\ Id., p. 116.
---------------------------------------------------------------------------
Insurance companies have long blamed liability laws for escalating
insurance premiums, and proffered restrictions on compensation to auto
accident victims--so-called ``no fault'' laws--as the solution. ``No
fault'' barred or gravely limited compensation to people for so-called
non-economic losses: principally the intangible pain and suffering
uniquely experienced by a human being that cannot be reduced to a
specific dollar value. In exchange, the insurance industry promised
lower premiums and richer insurance benefits for objective out-of-
pocket losses such as medical expenses and wage loss. At its peak,
twenty-four states had adopted some form of ``no fault'' auto
insurance.
As a practical matter, however, ``no fault'' proved to be a
disaster for consumers. ``No fault'' auto insurance became vastly more
expensive than the traditional liability system, and insurance
companies quickly argued they needed to cut the benefits in order to
bring prices under control.\89\
---------------------------------------------------------------------------
\89\ H. Rosenfield, ``Auto Insurance: Crisis and Reform,'' 29 U.
Memphis Law Review 69 (Fall 1998), p. 87-97.
---------------------------------------------------------------------------
The turning point was the electoral contest over insurance reform
in California. The insurance industry and its allies placed two ``no
fault'' related proposals on the California ballot in 1988, as an
alternative to Proposition 103. They were rejected by the voters by a
three to one margin.\90\ Insurance companies placed another ``no
fault'' initiative on the ballot in 1996. It, too, was decisively
rejected, with 65 percent of Californians voting against it.\91\ Four
states significantly altered or repealed their no-fault systems between
1989 and 1995: Georgia, Connecticut, Pennsylvania, and New Jersey,
experiencing rate reductions as a result.\92\ Today, only twelve states
employ any form of ``no fault'' insurance.
---------------------------------------------------------------------------
\90\ Id., p. 83-84.
\91\See https://ballotpedia.org/
California_Proposition_200,_No_Fault_Automobile_Insurance_
(1996) (last visited June 12, 2017).
\92\ H. Rosenfield, ``Auto Insurance: Crisis and Reform,'' 29 U.
Memphis Law Review 69 (Fall 1998), p. 87-97.
---------------------------------------------------------------------------
4.3 Repealing protections against insurance company price gouging and
discrimination
The insurance industry reliably opposes any form of regulation or
consumer protection legislation, and the potentially destabilizing
advent of self driving vehicles, with its host of unique and
unprecedented risks to consumers, is certain to inspire the
consideration of broader regulation at the state level. Proposition
103's protections will no doubt be considered a model for consumers in
other states as automated vehicles are rolled out.
Insurance companies vehemently opposed Proposition 103 at the
ballot box--spending a record $63 million in their campaign to defeat
it--and many insurers have sought to evade or contest its reforms since
they were upheld in a series of unanimous decisions by the California
Supreme Court after the measure passed. The industry, as well as
individual insurance companies, continues to fight the rate reductions
and premium rollbacks in the courts.\93\ So it is hardly a surprise
that some insurance companies hope to exploit the discussion about
insurance and liability in the era of autonomous vehicles to argue
that, as one industry source candidly put it, ``the Prop 103 model
should be scrapped entirely.'' \94\
---------------------------------------------------------------------------
\93\ ``CA Supreme Court Rebuffs Insurance Industry Assault on
Proposition 103 Rate Protections,'' Consumer Watchdog, May 11, 2017
(http://www.consumerwatchdog.org/newsrelease/ca-supreme-court-rebuffs-
insurance-industry-assault-proposition-103-rate-protections).
\94\ I. Adams, ``Does Prop 103 Violate Itself?'' Personal Insurance
Federation of California, October 21, 2014 (http://www.pifc.org/prop-
103-violate/).
---------------------------------------------------------------------------
The industry contends that Proposition 103's protections against
discriminatory rates and practices are outdated and will no longer be
necessary once robots, not humans, are driving vehicles.\95\ However,
as discussed above, no fully autonomous vehicle is available for
purchase today, nor has any date been set for the sale of such
vehicles, and America is decades away from a fully autonomous
transportation system (if it ever happens). Between now and that very
distant future, our roads will be a ``Hybrid Highway'' of vehicles with
greatly varying degrees of automation, ranging from none to a great
deal. So long as consumers are subject to liability for injuries and
property damage caused by the crash of a self-driving car or truck,
they will require insurance coverage. And so long as insurance
companies attempt to overcharge motorists for that protection, the
protective provisions of Proposition 103 will remain essential.
---------------------------------------------------------------------------
\95\ Id. See also R. Peterson, New Technology--Old Law: Autonomous
Vehicles and California's Insurance Framework, Santa Clara University
School of Law, May 21, 2012; D. Jergler, ``Prop. 103 vs. Self-driving
Cars Revving up in California,'' Insurance Journal, Sept. 17, 2014
(http://www.insurancejournal.com/news/west/2014/09/17/340898.htm); I.
Adams, ``Can Prop 103 Handle Driverless Cars?'' R Street, August 27,
2014, (http://www.rstreet.org/2014/08/27/can-prop-103-handle-
driverless-cars/).
---------------------------------------------------------------------------
Self driving vehicles will place the insurance industry at a
crossroads. Rather than resist or work to undermine reform, insurance
companies would be better advised to focus their resources on the
extremely important consumer protection role they could choose to play
as vehicle automation increases. Historically, the insurance industry
has exhibited limited interest in safety and ``loss prevention,''
perhaps because insurers are cost-plus, cash flow based institutions:
their profits are largely based on their projected costs, so when
claims rise, insurers can justify charging higher premiums, and earn
more investment income.\96\ These incentives have discouraged insurance
companies from using their vast information database on vehicle hazards
to alert manufacturers of vehicle dangers and press them--and
lawmakers--for safety improvements. This moment in history, marking a
rapid evolution in vehicle technology, is the time for the insurance
industry to weigh in--with a commitment to strong Federal safety
regulation, for example, and much more resources for affiliated
organizations whose mission is public safety and loss prevention.
---------------------------------------------------------------------------
\96\ H. Rosenfield, ``Auto Insurance: Crisis and Reform,'' 29 U.
Memphis Law Review 69 (Fall 1998), p. 76.
---------------------------------------------------------------------------
5.0 Guiding Principles
To protect consumers against the challenges posed by autonomous
vehicle technology, Consumer
Watchdog believes six principles must be adopted.
1. Protect the civil justice system. The state-based civil justice
system--open courts, impartial judges and citizen juries--is fully
equipped to handle the determination of legal responsibility as our
transportation system evolves over the coming decades. Disputes over
who is at fault in a crash involving a self-driving car or truck will
require the full power of civil justice system, with its procedural
safeguards of an impartial judge, full public transparency, and trial
by citizen juries, to investigate and publicly expose the cause of
crashes, compensate the victims for deaths, injuries and property
damage, punish the wrongdoer, and force manufacturers to make changes
in their products to prevent future harm. When their autonomous
technologies fail, hardware and software manufacturers must be held
strictly liable. Lawmakers should reject legislation to limit or
restrict state consumer protection laws. Manufacturers must not be
permitted to evade these consumer protections by inserting arbitration
clauses, ``hold harmless'' provisions or other waivers in their
contracts.
2. Enact stronger state consumer protections against insurance
company abuses. According to a 2013 report by the Consumer Federation
of America, ``California stands out from all other states in having the
best regulatory system for protecting consumers.'' Enacted by
California voters in 1988, California's insurance reform law provides
precisely the stronger protections consumers will require in the era of
robot vehicles. The reforms, known as Proposition 103, have protected
motorists (along with homeowners, renters, businesses and medical
providers) against unjust insurance rates (including product liability
insurance rates) and anti-consumer and discriminatory practices. The
law's emphasis on rewarding drivers with lower insurance premiums based
on their safety record, their annual mileage, their driving experience,
and other rating factors within their control that are ``substantially
related to the risk of loss,'' will be critical in the new automotive
era. Proposition 103's mandate for public disclosure and public
participation in regulatory matters are essential components of a
system that will be trusted by consumers.
3. Enact auto safety standards. Private companies cannot be trusted
to develop and deploy robot cars and trucks without rules. The Federal
auto safety agency, or in its absence, state auto safety agencies, must
develop standards for the testing and deployment of the multiple
technologies required by robot vehicles. These standards must address
safety; security; privacy and the software that determines the robot's
actions in the event of an impending collision and as it makes life and
death decisions. They must be enforceable by consumers in courts of
law.
4. Stronger laws are needed to protect consumers' privacy. The laws
have not kept pace with the evolution of technology and the collection
and monetization of consumers' personal data. Hardware and software
manufacturers and insurance companies must be barred from utilizing
tracking, sensor or communications data, or transferring it to third
parties for commercial gain, absent separate written consent (which
should not be required as a condition of accessing the services of the
vehicle/manufacturer, and which should be revocable by the consumer at
any time).
5. Bar Federal interference in state consumer protection laws.
Neither Congress not Federal agencies should be permitted to preempt or
override stronger state based civil justice, insurance reform or auto
safety laws.
6. Respect democratic and human values. The sponsors of self-
driving vehicles have promoted the myth that machines are infallible in
order to justify the wholesale departure from a panoply of norms that
form founding principles for the nation, beginning with the rule of
law; individual and corporate responsibility; long held legal
principles that distinguish between human beings and property; and the
transparency of public officials and institutions that is a hallmark of
democracy. The strategy of substituting robot values for human values
has reached its apotheosis in the determination by robot car company
executives to program computers to make life and death decisions, and
to keep that decision-making process secret. Lawmakers will need to
impose the rule of law and other attributes of American democracy upon
the executives of the hardware and software companies that manufacture
self-driving cars.
______
Response to Written Question Submitted by Hon. Brian Schatz to
Mitch Bainwol
Question. There is conflicting research on whether self-driving
cars will increase congestion and vehicle emissions overall. For
instance, vehicle-miles traveled could increase because automation
lowers the opportunity cost of driving. Do you think the widespread
deployment of self-driving cars will increase congestion or vehicle
emissions? How do we ensure that this does not happen?
Answer. It is thought that self-driving vehicles will be initially
deployed primarily in a ride-hailing, ride-sharing, or goods delivery
fleet context. These vehicles will be part of Transportation as an
Operating ecosystem that will allow better fleet management and more
efficient transportation, which, coupled with the general trend toward
drivetrain electrification, offers the possibility of reduced
congestion and emissions. We are also excited about potential for self-
driving vehicles to reduce traffic fatalities, improve transportation
efficiency, and expand mobility to traditionally underserved
populations. We feel those reasons are quite compelling in the overall
debate about permitting the deployment of self-driving vehicles.
Regardless of vehicle powertrain, it would be speculative to
definitively state that self-driving vehicle deployment will increase
vehicle emissions or congestion. However, it is well documented that
current advanced driver assist technologies have a positive impact on
congestion because the technology is preventing accidents that would
otherwise occur due to driver error.
Note, attached is a projection from Moody's Investors Services that
outlines the expected rate of use and ultimate saturation for HAV
technologies, including fully autonomous vehicles.
______
Response to Written Question Submitted by Hon. Tammy Duckworth to
Rob Csongor
Question. Data access is an issue raised by a number of
stakeholders, ranging from public safety advocates, law enforcement,
insurers, and environmental organizations. Considering the importance
of data access for liability determinations, environmental impact
determinations, and determining how safe this technology actually is,
how will these issues be addressed to make sure the appropriate
stakeholders have access to relevant data that is not proprietary
information?
Answer. In order to develop self-driving cars and ensure they are
safe, massive amounts of data are required to train AI neural networks.
In addition, we will want to test them on real roads as well as
simulated tests that will be based on data from previously driven
miles.
All of this data can be collected from fleets of test vehicles, as
well as customer vehicles. NVIDIA believes that the data obtained by
customer vehicles must be treated in a fashion that anonymizes the
data. No personally identifiable information should be retained, but
rather just the sensor data and other vehicle data will be maintained
in order that the driving system can be trained.
With respect to archiving data on the vehicle, it makes sense to
have a data recorder that maintains recent driving activity that can be
used to understand what happened in the case of an incident, and can be
used to determine who is at fault.
______
Response to Written Questions Submitted by Hon. Ron Johnson to
John M. Maddox
Question 1. A recent Deloitte research paper illustrates how the
transportation industry is leveraging the wireless platform to innovate
and grow. According to this paper, wirelessly connected self-driving
cars could reduce travel times by nearly 40 percent and delays by 20
percent, annually generating $447 billion in savings and saving 21,700
lives. Can you describe the American Center for Mobility's work with
wireless providers to speed up the delivery of self-driving vehicles?
Answer. Our current transportation system is comprised of people,
vehicles, and roads (infrastructure). The American Center for Mobility
(ACM) is convinced that modern communications networks will make up a
critical and necessary component of our future transportation system,
right alongside people, vehicle, and roads. This modern transportation
communications network will enable the rapid distribution and
collection of data, forming a data backbone which will create the
possibility for a systems-approach for the efficient flow of people and
goods.
Highly Automated Vehicles (HAVs) will rely heavily on this data
backbone. HAV technology is critical to a future transportation system
that continues to support our national economy, and deployment of these
communications and vehicle technologies will help ensure our continued
international economic competitiveness.
ACM is creating a national-scale proving ground for the future of
transportation where these future transportation products and services
can be tested, verified, and validated. In addition to testing for
vehicles and roads, ACM is building a facility for critical testing and
development of the communications systems that will form this data
backbone.
ACM is collaborating with AT&T, and other companies who will be
announced at a later date, to design, build, and operate a dedicated
network so that manufacturers, developers, service providers,
government bodies, and other stakeholders can rapidly develop and
deploy beneficial transportation technologies in a collaborative
fashion. This dedicated network will include, among other technologies,
4G/LTE cellular, Dedicated Short Range Communication (DSRC), Wi-Fi,
cloud services, edge computing, and 5G when it is ready.
We anticipate that this transportation communications testing
capability will be in high-demand and will enable the very rapid
development of Connected and Automated Vehicle technology.
Question 2. As more connected and autonomous vehicles hit the road,
they will need to ``talk'' to each other and everything around them in
a secure manner to realize the full potential of the technology. There
will be an enormous growth in data as vehicles and surrounding
infrastructure become connected. How is the American Center for
Mobility working with the wireless industry to ensure that the wireless
infrastructure exists to handle the vast amounts of data needs that
will come with autonomous vehicles?
Answer. The American Center for Mobility (ACM) is convinced that a
modern communications network is a required component of our future
transportation system to enable needed data systems. Highly Automated
Vehicles (HAVs) present both challenges and opportunities regarding
data and data sharing, as the vehicles themselves require that a large
amount of data and information be sensed, acquired, amalgamated,
analyzed for rapid decision-making, and acted upon through control
decisions and operational monitoring. However, the processing,
distribution, and eventual storage of these extremely large amounts of
data is itself a significant challenge. It is highly unlikely that it
will be practical to transfer and store all of the data from thousands,
much less hundreds of thousands, of vehicles throughout the vehicle
lifetime. Therefore, data selection and harvesting processes must be
developed.
ACM is working with telecom, automotive, cybersecurity, computing,
simulation, and infrastructure industry partners to create a model
communications system whereby they can test, develop, verify, and
validate their key technology and offerings.
We are also focused on partnering with key Standards Development
Organizations (SDOs) to rapidly create, publish, and update critical
voluntary standards around communications systems, as well as the
vehicle and roads themselves. It is critical that we begin and
accelerate the work to establish these voluntary standards to support
the deployment on this technology and to help ensure our continued
international economic competitiveness.
______
Response to Written Questions Submitted by Hon. Brian Schatz to
John M. Maddox
Question 1. It is reported that driving a self-driving car for an
hour will generate four terabytes of data. This data is very valuable
for various applications--urban planning, for example. But there are
currently a wide variety of approaches in state legislations when it
comes to data ownership. What do you think are good principles of data
sharing among car operators, car manufacturers, and the government?
Answer. Highly Automated Vehicles (HAVs) present both challenges
and opportunities regarding data and data sharing, as the vehicles
themselves require that a large amount of data and information be
sensed, acquired, amalgamated, analyzed for rapid decision-making, and
acted upon through control decisions and operational monitoring.
However, the processing, distribution, and eventual storage of these
extremely large amounts of data is itself a significant challenge. It
is highly unlikely that it will be practical to transfer and store all
of the data from thousands, much less hundreds of thousands, of
vehicles throughout the vehicle lifetime. Therefore, data selection and
harvesting processes must be developed.
While data best-practices for HAVs are still under significant
development by vehicle manufacturers, testers, suppliers, and service
operators, generally, these data practices will eventually likely
include established processes for maintaining privacy and proper
ownership, while providing for the collection of in-use event,
incident, and vehicle information data for crashes, near crashes,
malfunctions, degradations, failures, and unintended operation outside
established operational domains. It is expected that this information
will become extremely useful for vehicle development, safety
monitoring, and operations activities as well as accident/event
reconstruction purposes. All of these types of data can be useful for
creating appropriate safety requirements, but privacy and ownership
principles must be maintained.
Key principles must be established at the policy level, ideally
federally to avoid a patchwork of requirements, and in concert with
industry and consumer groups. The basic principles employed in today's
vehicles are appropriate starting points, including the concept that
the vehicle-level data ownership is dictated by vehicle ownership,
unless voluntarily sharing is ``opted-into'' by the owner.
Lastly, data ownership and sharing policies must also take into
account the reality that cellular phones, tablets, wearables, and other
devices are already collecting and transmitting large amounts and broad
types of data, and that vehicle-based data is not the only source of
privacy and ownership concerns in the transportation system.
Question 2. This year, there has been a lot discussion from both
sides of the aisle on our Nation's deteriorating infrastructure. In
your opinion, is our physical infrastructure ready for self-driving
cars? If not, what has to change?
Answer. Our transportation system is the lifeblood of the U.S.
economy. It supports all of our business and economic activities,
including the basics of getting people to their jobs and goods to
market. It underpins our society and our entire way of life. Our
deteriorating infrastructure is already weighing down our economy with
increased congestion, wasted fuel, and significant safety issues.
Further lack of investment in our infrastructure will also hamper our
ability to deploy Highly Automated Vehicle (HAV) technology in the long
term at large scale and will reduce our international economic
competitiveness.
While our current infrastructure is less than ideal, HAVs will
initially be able to accommodate this, much like initial automobiles
handled horse-trails 100 years ago. But quickly our existing
infrastructure, if left unimproved, will increasingly inhibit
widespread deployment of HAV technology and significantly reduce
societal benefit.
In the short term, we should concentrate on improving the basics,
including well-painted lane markings, high quality and consistent
signage, and well-lit and delineated pedestrian crossings and bicycle
lanes. We should move quickly to deploy Vehicle-to-Infrastructure (V2I)
and Vehicle-to-Vehicle (V2V) communications systems based on DSRC and
cellular, as well as regularly updated detailed maps to provide HAVs
with augmented data for safety and operations.
In the longer term, we should begin to establish best practices for
new infrastructure designs for urban and rural areas, including managed
lanes, further improved lane and road edge markings, traffic control
systems, and parking.
______
Response to Written Question Submitted by Hon. Tammy Duckworth to
John M. Maddox
Question. Auto manufacturers and technology companies continue to
develop and push for highly autonomous vehicles to enter our
transportation landscape and we understand that in order to bring these
products to market they need to be tested not only on secure test
tracks but also on actual roads. Many of these vehicles under
development are being designed to have little if any human intervention
features should something arise that requires someone or something to
take control of the vehicle.
Is the role of insurance being addressed as autonomous vehicle
technology is developed and how do you envision liability issues
(product liability, strict liability, etc.) going to be addressed going
forward as we move toward the Level 3-5 environment?
Answer. It is true that numerous companies are developing highly
automated vehicles for deployment and operation over the next 1-5
years. Some of these vehicles will not require human intervention for
normal operation. And indeed, best practices for development and full
validation of these vehicles include a coordinated and diligent
approach including both on-road and closed course testing, as well as a
related computer-simulation approach.
The role, and nature, of insurance is being considered and
logically must be addressed in this same time period. Currently,
insurance companies and re-insurance companies are analyzing this
technology as they would with any new technology: examining, collecting
data where possible, and attempting to quantify the risk and potential
benefit of the technology. However, it is largely understood that data
is scarce and that the examination is still in early stages.
Predictably, insurance companies are generally not offering a public
view of their planning and analysis efforts or results.
The assumption in the vehicle developer community is that the
current liability system in the U.S. will likely be in place throughout
the near-term roll-out of the technology. However other countries and
regions, including those with highly developed automotive industries,
are critically examining whether other liability regimes, including
shared liability, would provide greater societal benefit due to faster
deployment or other means.
[all]