THE REASONABLE SELF-DRIVING CAR
it performs. These arguments are central to functional safety
By Bryant Walker Smith on October 3, 2013 at 2:38 am
Cross-posted from Volokh Conspiracy.
A common debate in many circlesincluding the comments on my
posts hereis whether legal burdens, technical limitations, or
consumer preferences present the greatest immediate obstacle to fully
automated motor vehicles. (Fully automated vehicles are capable of
driving themselves anywhere a human can. In contrast, the low-speed
shuttles from my post on Monday are route-restricted, and the
research vehicles that regularly appear in the news are both route-
restricted and carefully monitored by safety drivers.)
An entirely correct response is that the technologies necessary for full
automation are simply not ready. Engineering challenges will be
overcome eventually, but at this point they are varied and very real. If
they were not, we would already see fully self-driving cars
operating somewhere in our diverse worldin Shanghai or Singapore,
Abu Dhabi or Auckland.
The deeper issue, which manifests itself in law, engineering, and
economics, is our (imperfect and inconsistent) societal view of what
is reasonably safe, because it is this view that determines when a
technology is ready in a meaningful sense. Responsible engineers
will not approve, responsible companies will not market, responsible
regulators will not tolerate, and responsible consumers will not
operate vehicles they believe could pose an unreasonable risk to
safety.
How safe is safe enough? One answer, that self-driving cars must
perform better than human drivers on average, accepts some deaths
and injuries that a human could have avoided. Another answer, that
self-driving cars must perform at least as well as a perfect human
driver for every individual driving maneuver, rejects technologies
that, while not perfect, could nonetheless reduce total deaths and
injuries. A third answer, that self-driving cars must perform at least as
well as corresponding human-vehicle systems, could lock humans
into monitoring their machinesa task at which even highly trained
airline pilots can occasionally fail due to understimulation or
overstimulation.
The secondary effects of automated vehicle crashes challenge these
answers. A particularly tragic, sensational, or unusual crash could
ultimately claim more lives by tarnishing technologies that might
nonetheless represent a safety gain over human driving. (In other
words, headlines about a single self-driving car crash could
trump 30,000obituaries.) Conversely, early incidents could ultimately
save lives by providing the real-world data needed to accelerate the
design of even safer systems.
Demonstrating reasonable safety may be more difficult than defining
it. A rough non-Bayesian statistical calculation suggests that
a fully automated vehicle concept would need to accumulate over
700,000 miles of unassisted driving in representative conditions to
establish with 99 percent confidence that it crashes less frequently
than conventional cars. An international standard for functional safety
similarly establishes failure rates so low that showing that a system
meets the least restrictive level would involve testing the system
continuously for more than ten years, under operational conditions,
with no unsafe failures and no modifications to it (PDF). (An effort
to adapt the automotive-specific standard (ISO 26262) to automated
driving is ongoingbut will be for a very long time.)
These big numbers mean that an engineering safety case may need to
rely on evidence beyond just empirical testing. From a legal
perspective, I am particularly interested in, and would especially
welcome your thoughts on, the role of process-based safety
arguments, which focus on how a product is designed rather than how
standards, including ISO 26262. They also implicate the tension in
law, as in engineering (PDF), between processes (or inputs) and
products (or outputs).
In tort law, negligence is about process (how did the manufacturer
perform?) while strict liability is about product (how did the
manufactured item perform?), and a reasonably safe process can
occasionally produce an unreasonably safe product. As sporadic
failures of automated vehicles inevitably occur, negligence claims,
punitive damage awards, and determinations of foreseeability may all
depend in part on the reasonableness of a defendant manufacturers
prior process-based safety arguments. (Much more has already been
written on liability, and I will add another perspective tomorrow.)
In administrative law, regulation of outputs (how fast must a car be
able to stop?) is generally preferable to regulation of inputs (what
kind of brakes must a car have?). Because of the difficulty in
prospectively defining and demonstrating automotive safety,
however, initial regulatory efforts may need to emphasize inputs over
outputs. For example, a state or federal agency might require
automated vehicle developers to provide persuasive evidence of their
engineering competence, safety record, and financial solvency before
publicly testing or marketing their vehicles. Because agencies will
need broad discretion to experiment, to adjust, and to impose ad hoc
requirements that may border on the arbitrary, judicial deference to
inchoate agency practice is especially important.
Such an input-based regulatory approach might impede start-ups and
other small actors from independently developing or marketing
automated vehicle technologies. Indeed, Nevada already imposes
barriers to testing that could have this effect. This may be
undesirable, particularly if rapid innovation is the paramount goal.
However, it would mean that established companies with significant
financial and reputational interests would likely be the first ones to
vouch for the reasonable safety of these systems. Given the enormous
stakes, indirectly forcing this kind of deliberation may be prudent.
SELF-DRIVING CARS WON'T JUST WATCH THE WORLD
THEY'LL WATCH YOU
I T S M O N D AY M O R N I N G, youre late for work, and as you
merge onto the freeway you see it: the sea of red brake lights. Its
going to be a slow, frustrating tripfor all the suckers who have to
drive their own cars. You click yours into autonomous mode and
spend the slog getting ahead on work emails, or even catching up on
sleep.
Yes, the day you become a co-driver is fast approaching. But as cars
master how to see, understand, and navigate the world, researchers
are shifting their attention to another subject: you. Paradoxical it may
seem, but the more control the car has, the more it needs to know
about the person sitting behind the wheelwhether they're paying
attention, their mood, even their health.
We are making tremendous progress in instrumenting vehicles to
know everything thats happening around them, but there are just not
enough sensors looking at the driver inside the car, says Anuj
Pradhan, who studies human factors at the University of Michigans
Transportation Research Institute.
Used to be, if you stopped paying attention while driving, you'd just
crash. And in 20 or 50 years, when cars are 100 percent autonomous,
whatever you're up to won't matter, because you'll have zero
responsibility. Today's technology sits between those points: The
robots are doing some of the work. Tesla already sells cars that drive
themselves on the highway, as long as the human monitors the
system, ready to take over at any moment. Next year, Audi plans to
introduce a more capable system, where the driver is demoted from
supervisor to understudy, necessary only when things go to pot.
A lot of the players in this business hate that idea (a bunch are
avoiding that kind of system) because people are godawful backups.
They're prone to dozing off, zoning out, goofing around. But if you
want an autonomous car that can roam beyond a constrained
geographical zone, or that can stay on the road in less than ideal
weather conditionsand you want it this decadeyou're gonna need
some human help.
Distractions Can Be a Good Thing
So researchers and engineers in the autonomous space are focusing
more and more attention on the human. One surprise: Smartphones
can help. Being in an autonomous car is incredibly boring, and we
have a lot of people who fall asleep, says Wendy Ju, who studies
self-driving cars at Stanfords Center for Design Research.
Distractions like texting and tweeting, super dangerous in a regular
car, can be useful in a self-driving one, engaging the human's brain.
Demanding a human take the wheel is way harder when that person's
sleeping. These are things that keep you awake," Ju says. "Theyre
actually good.
Greatas long as car knows what the human's up to, and whether
they're able to take control of the car if needed. Basic driver
monitoring systems have been around for more than a decade, mostly
aimed at combatting drowsy driving. In 2003, Volvo introduced its
Intelligent Driver Information System, which monitors steering wheel
and pedal inputs, and whether the turn signal is on. That's enough to
guess if the driver's in the middle of a high stress overtaking
maneuverand it's better to automatically decline that incoming call.
Some BMW models will pop up an icon of a steaming cup of coffee
if steering inputs start wandering, and it seems like the driver could
be nodding off. Toyota has used a camera to watch the driver's
eyelids.
This new challenge invites a more Orwellian approach. Australian
company Seeing Machines says its gaze tracking technology will
allow cars to act as co-drivers, because they'll know what the driver
has and hasn't seen. Industry supplier Pioneer wants to monitor his
heart rate. Harman is working on tech that measures pupil dilation,
aiming to understand cognitive load.
Your Car Will Pick up on Your Moods
This richness of information is likely crucial for the semi-autonomous
car, but it could also inform how truly driverless systems work. Think
about when youre a passenger in the front seat of a regular car, and
the driver is speeding, or changing lanes erratically. You may tense
up, frown, tug at your seatbeltcommunicating youd prefer to slow
down, please, without having to say it.
Even if you're happy with the driving style, you may point out things
the driver missed, or suggest a different route. These could all be
possible with an autonomous car too, with the right sensors pointed at
the person in what was the drivers seat.
This goes both ways. The cars computer can better calibrate how it
talks to the driver, like with a louder warning if he is obviously
distracted. It could also offer to engage autonomous systems if the
road conditions look good and the driver looks tired.
Having a camera pointed at your face raises obvious privacy
concerns, but Ju says its unlikely all that data will be collected and
kept. It would be very expensive from a bandwidth perspective to
transmit video of what youre doing in the car. And if that changes,
it may just be the price you pay for improved safety. Until your
autonomous car can cope without a human all together, and you
become, at long last, irrelevant.
CONGRESS COULD MAKE SELF-DRIVING CARS
HAPPENOR RUIN EVERYTHING
C O N GR E S S J U S T S TE P PE D into the robocar game. In the
past two days, a pair of senators started drafting legislation to
advance autonomous vehicles, and the House Subcommittee on
Digital Commerce and Consumer Protection held a two-hour hearing
exploring how on the tech might be deployed. For your elected
officials, it's a considerable, if tentative, step into the future of
transportation.
Of course, they're just a bit late. Small numbers of robocars already
roam the San Francisco Bay Area and other cities, and you'll
probably start riding in them within a few years as Uber and others
commercialize the technology. Everything is racing ahead of a
regulatory structure ill-equipped to usher in this change.
The nation's patchwork of laws regulating this technology say
nothing about how it is tested (or even defined), how cars using it
will operate, or even who should settle these questions. Congress can
address these all of these questions and ensure this technology
succeeds.
Or they can screw it all up.
"I'd be wary of dramatic proposals that could create more problems
than they solve," says Bryant Walker Smith, an expert on
autonomous vehicles at the University of South Carolina School of
Law. After all, this evolving technology permeates so many parts of
society: public safety, privacy, the environment, liability and
insurance law, employment, urban planning, and more. A law aimed
at cutting congestion could tangle with tort law; a clause ensuring
passenger privacy could wipe out economic benefits for automakers.
State legislators have already stepped in it. The folks behind
California's autonomous vehicles law couldn't even define
"autonomous vehicle" without giving Uber a plausible argument for
ignoring the rules. Michigan's law favors established automakers over
newcomers like Google and Uber. Nevada managed to write a law
requiring anyone operating an autonomous vehicles to change its
license plate upon entering the state.
New Law for a New Era
Automakers and tech companies developing autonomous technology
live in fear of states drafting their own rules, creating a patchwork of
regulations. That's exactly what's happening, because
although federal regulators dictate almost everything about a car you
can think of (brake lights must be red, turn signal icons are green,
passenger cars have airbags), states regulators stipulate their
operation by setting speed limits, traffic laws, and so on.
That's why the GMs and Googles of the world see federal
intervention as a potential savior. Federal oversight provides a broad,
consistent framework for testing and deploying their robots. During
Tuesday's House subcommittee hearing, representatives of General
Motors, Toyota, Volvo, and Lyft said they're OK with the voluntary
15-point checklist the federal Department of Transportation released
last fall. (Among other things, the guidelines favor greater flexibility
by federal regulators and give automakers wide latitude in how they
prove the safety of their technology.) The guidelines warn states
against creating additional regulations, and don't dive too deeply into
stipulating what kind of technology everyone usesTeslas
opposition to LIDAR is OK, as is Fords lack of interest in vehicle-
to-infrastructure tech and GM's dedication to deploying ride-sharing
robocars with Lyft.
The industry definitely wants the feds to step it up. Do you agree
that Federal Motor Vehicle Safety Standards need to be updated in
order to support the deployment of automated vehicles?
Representative Debbie Dingell, a Democrat from Michigan, asked
industry witnesses. One by one, they went down the row: Yes. Yes.
Yes. Yes. Yes.
Of course, they aren't the only stakeholders. Any major legislation
should consider the concerns of independent researchers, privacy and
consumer advocates, transportation experts, driver unions, and others.
Further complicating matters, the technology is so new that no one
agrees on how a finished product will work, or even how to
determine when a robocar is ready for primetime.
"We neither know what test autonomous vehicles should have to take
nor what should constitute a passing grade," testified Nidhi Kalra, co-
director and senior information scientist at the RAND Center for
Decision Making Under Uncertainty.
Nips & Tucks
If Congress lacks the desire to study the issue carefully and tackle it
with a comprehensive law, it can go at things piecemeal and still
nudge automation along. It could start by revising the Federal Motor
Vehicle Safety Standards to reflect autonomous technology. For
example, the rules require things like foot-activated brakes. The
National Highway Transportation Safety Administration can amend
the regulations, but it requires several rounds of draft rules and public
comments. That takes years. Congress can make the same change
quickly with a law, or even a clause tucked into, say, an infrastructure
omnibus.
It could also tweak Title 49 of the US legal code, which allows the
secretary of transportation to exempt vehicles from federal
standardsto a point. The law (specifically section 30113(d) and
30113(e)) limits those vehicles to 2,500 per manufacturer in a 12-
month period, and the exemption can't last more than three years.
Federal regulators may be happy to let Google produce a car without
a steering wheel or brakes, but without congressional help, that
goodwill can only go so far.
Congress also could give regulators the right to approve these
vehicles before they go on sale. The National Highway Traffic Safety
Administration springs into action when cars already on the
road prove themselves unsafe. When the DOT published its AV
guidelines last year, it suggested Congress treat self-driving cars like
aircraft, drugs, and medical devicesunsafe and un-OK until
evidence says otherwise. You could argue that will slow things down,
but ensuring these cars are safe before they hit the road could
avoid crashes that might set the whole industry back.
That one's a long shot, given that the industry hates the idea(you'll
stifle innovation!) and today's Republican-dominated government
hardly lusts after more regulation. "It doesnt look like theres a lot of
enthusiasm for that," says Smith, the legal expert. "And the agency
would need a lot more resources." Don't expect Congress to unlock
more funding, either.
Whatever Congress decides, it must chart a careful course. The age of
the robocar is nigh, and the transition will stir up some scary seas.
Aarian Marshall contributed reporting.
Self-Driving Cars Could Make Moral Decisions Like Humans
With a Simple Algorithm Deciding who lives and who dies.
GEMMA CONROY
6 JUL 2017
Self-driving cars are almost here, but one big question remains - how
do they make hard choices in a life and death situation? Now
researchers have demonstrated that smart vehicles are capable of
making ethical decisions on the road, just like we do everyday.
By studying human behaviour in a series of virtual reality-based
trials, the team were able to describe moral decision making in the
form of an algorithm. This is huge because previously, researchers
have assumed that modelling complex ethical choices is out of reach.
"But we found quite the opposite," says Leon Stfeld, one of the
researchers from the University of Osnabruck, Germany. "Human
behaviour in dilemma situations can be modelled by a rather simple
value-of-life-based model that is attributed by the participant to every
human, animal, or inanimate object."
If you take a quick glance at the statistics, humans can be pretty
terrible drivers that are often prone to distraction, road rage and drink
driving. It's no surprise then that there are almost 1.3 million
deaths on the road worldwide each year, with 93 percent of accidents
in the US caused by human error.
But is kicking back in the seat of a self-driving car really a safer
option? The outlook is promising. One report estimates that driverless
vehicles could reduce the number of road deaths by 90 percent, which
works out to be around 300,000 saved lives a decade in the US alone.
Despite the glowing figures, developing a self-driving car that can
respond to unpredictable situations on the road hasn't been a smooth
ride. One stumbling block is figuring out how these smart cars will
deal with road dilemmas that require ethical decision-making and
moral judgement.
While previous research has shown that self-driving cars can avoid
accidents by driving slow at all times or switching to a different
driving style, following a set of programmed 'rules' isn't enough to
survive the demands of inner city traffic.
And then there's the 'trolley dilemma', a thought experiment that tests
how we make moral decisions. In this no-win scenario, a trolley is
headed towards five unsuspecting bystanders. The only way to save
these five people is by pulling a lever that diverts the trolley onto
another track where it kills one person instead. How can a driverless
car make the best choice?
The tricky thing with these kinds of decisions is that we tend to make
a choice based on the context of the situation, which is difficult to
mirror in the form of an algorithm programmed into a machine.
With this in mind, Stfeld and the team took a closer look at whether
modelling human behaviour when driving is as impossible as
everyone thinks it is.
Using virtual reality to simulate a foggy road in a suburban setting,
the team placed a group of participants in the driver's seat in a car on
a two-lane road. A variety of paired obstacles, such as humans,
animals and objects, appeared on the virtual road. In each scenario,
the participants were forced to decide which obstacle to save and
which to run over.
Next, the researchers used these results to test three different models
predicting decision making. The first predicted that moral decisions
could be explained by a simple value-of-life model, a statistical term
measuring the benefits of preventing a death.
The second model assumed that the characteristics of each obstacle,
such as the age of a person, played a role in the decision making
process. Lastly, the third model predicted that the participants were
less likely to make an ethical choice when they had to respond
quickly.
After comparing the results of the analysis, the team found the first
model most accurately described the ethical choices of the
participants. This means that self-driving cars and other automated
machines can make human-like moral choices using a relatively
simple algorithm.
But before you throw away your driver's license, it's important to
remember that these findings open up a whole new realm of ethical
and moral debate that need to be considered before self-driving cars
are out on the road.
"Now that we know how to implement human ethical decisions into
machines we, as a society, are still left with a double dilemma," says
Peter Konig, one of the researchers on the study "Firstly, we have to
decide whether moral values should be included in guidelines for
machine behaviour, and if they are, should machines act just like
humans?"
While we still have a way to go before we take our hands off the
steering wheel for good, these findings are a big leap forward in the
world of intelligent machines.
AN ACT IN THE COUNCIL OF THE DISTRICT OF
COLUMBIA
To authorize autonomous vehicles to operate on District roadways, to
require the Department of Motor Vehicles to create an autonomous
vehicle designation, and to establish safe operating protocols for
autonomous vehicles.
Autonomous Vehicle Act of 2012.
Sec. 2. Definitions.
For the purposes of this act, the term:
(1) Autonomous vehicle means a vehicle capable of navigating
District roadways and interpreting traffic-control devices without a
driver actively operating any of the vehicles control systems. The
term autonomous vehicle excludes a motor vehicle enabled with
active safety systems or driver- assistance systems, including systems
to provide electronic blind-spot assistance, crash avoidance,
emergency braking, parking assistance, adaptive cruise control, lane-
keep assistance, lane-departure warning, or traffic jam and queuing
assistance, unless the system alone or in combination with other
systems enables the vehicle on which the technology is installed to
drive without active control or monitoring by a human operator.
(2) Driver means a human operator of a motor vehicle with a valid
drivers license.
(3) Public roadway means a street, road, or public thoroughfare
that allows motor vehicles.
(4) Traffic control device means a traffic signal, traffic sign,
electronic traffic sign, pavement marking, or other sign, device, or
apparatus designed and installed to direct moving traffic.
Sec. 3. Autonomous vehicles permitted.
An autonomous vehicle may operate on a public roadway; provided,
that the vehicle:
(1) Has a manual override feature that allows a driver to assume
control of the autonomous vehicle at any time;
(2) Has a driver seated in the control seat of the vehicle while in
operation who is prepared to take control of the autonomous vehicle
at any moment; and
(3) Is capable of operating in compliance with the Districts
applicable traffic laws and motor vehicle laws and traffic control
devices.
Sec. 4. Vehicle conversion; limited liability of original manufacturer.
(a) The original manufacturer of a vehicle converted by a third party
into an autonomous vehicle shall not be liable in any action resulting
from a vehicle defect caused by the conversion of the vehicle, or by
equipment installed by the converter, unless the alleged defect was
present in the vehicle as originally manufactured.
(b) The conversion of vehicles to autonomous vehicles shall be
limited to model years 2009 or later or vehicles built within 4 years of
conversion, whichever vehicle is newer.
Sec. 5. Rules.
The Mayor, pursuant to Title 1 of the District of Columbia
Administrative Procedure Act, approved October 21, 1968 (82 Stat.
1204; D.C. Official Code 2-501 et seq.), shall issue rules on or
before December 31, 2013, establishing a class of vehicles for
autonomous vehicles and procedures and fees for the registration,
titling, and issuance of permits to operate autonomous vehicles.
Sec. 6. Fiscal impact statement.
The Council adopts the fiscal impact statement in the committee
report as the fiscal impact statement required by section.
CA Veh. Sec. 38750 (California Code (2014 Edition))
38750. (a) For purposes of this division, the following definitions
apply:
(1) "Autonomous technology" means technology that has the
capability to drive a vehicle without the active physical control or
monitoring by a human operator.
(2) (A) "Autonomous vehicle" means any vehicle equipped with
autonomous technology that has been integrated into that vehicle.
(B) An autonomous vehicle does not include a vehicle that is
equipped with one or more collision avoidance systems, including,
but not limited to, electronic blind spot assistance, automated
emergency braking systems, park assist, adaptive cruise control, lane
keep assist, lane departure warning, traffic jam and queuing assist, or
other similar systems that enhance safety or provide driver assistance,
but are not capable, collectively or singularly, of driving the vehicle
without the active control or monitoring of a human operator.
(3) "Department" means the Department of Motor Vehicles.
(4) An "operator" of an autonomous vehicle is the person who is
seated in the driver's seat, or if there is no person in the driver's seat,
causes the autonomous technology to engage.
(5) A "manufacturer" of autonomous technology is the person as
defined in Section 470 that originally manufactures a vehicle and
equips autonomous technology on the originally completed vehicle
or, in the case of a vehicle not originally equipped with autonomous
technology by the vehicle manufacturer, the person that modifies the
vehicle by installing autonomous technology to convert it to an
autonomous vehicle after the vehicle was originally manufactured.
(b) An autonomous vehicle may be operated on public roads for
testing purposes by a driver who possesses the proper class of license
for the type of vehicle being operated if all of the following
requirements are met:
(1) The autonomous vehicle is being operated on roads in this state
solely by employees, contractors, or other persons designated by the
manufacturer of the autonomous technology.
(2) The driver shall be seated in the driver's seat, monitoring the
safe operation of the autonomous vehicle, and capable of taking over
immediate manual control of the autonomous vehicle in the event of
an autonomous technology failure or other emergency.
(3) Prior to the start of testing in this state, the manufacturer
performing the testing shall obtain an instrument of insurance, surety
bond, or proof of self-insurance in the amount of five million dollars
($5,000,000), and shall provide evidence of the insurance, surety
bond, or self-insurance to the department in the form and manner
required by the department pursuant to the regulations adopted
pursuant to subdivision (d).
(c) Except as provided in subdivision (b), an autonomous vehicle
shall not be operated on public roads until the manufacturer submits
an application to the department, and that application is approved by
the department pursuant to the regulations adopted pursuant to
subdivision (d). The application shall contain, at a minimum, all of
the following certifications:
(1) A certification by the manufacturer that the autonomous
technology satisfies all of the following requirements:
(A) The autonomous vehicle has a mechanism to engage and
disengage the autonomous technology that is easily accessible to the
operator.
(B) The autonomous vehicle has a visual indicator inside the cabin
to indicate when the autonomous technology is engaged.
(C) The autonomous vehicle has a system to safely alert the
operator if an autonomous technology failure is detected while the
autonomous technology is engaged, and when an alert is given, the
system shall do either of the following:
(i) Require the operator to take control of the autonomous vehicle.
(ii) If the operator does not or is unable to take control of the
autonomous vehicle, the autonomous vehicle shall be capable of
coming to a complete stop.
(D) The autonomous vehicle shall allow the operator to take
control in multiple manners, including, without limitation, through
the use of the brake, the accelerator pedal, or the steering wheel, and
it shall alert the operator that the autonomous technology has been
disengaged.
(E) The autonomous vehicle's autonomous technology meets
Federal Motor Vehicle Safety Standards for the vehicle's model year
and all other applicable safety standards and performance
requirements set forth in state and federal law and the regulations
promulgated pursuant to those laws.
(F) The autonomous technology does not make inoperative any
Federal Motor Vehicle Safety Standards for the vehicle's model year
and all other applicable safety standards and performance
requirements set forth in state and federal law and the regulations
promulgated pursuant to those laws.
(G) The autonomous vehicle has a separate mechanism, in addition
to, and separate from, any other mechanism required by law, to
capture and store the autonomous technology sensor data for at least
30 seconds before a collision occurs between the autonomous vehicle
and another vehicle, object, or natural person while the vehicle is
operating in autonomous mode. The autonomous technology sensor
data shall be captured and stored in a read-only format by the
mechanism so that the data is retained until extracted from the
mechanism by an external device capable of downloading and storing
the data. The data shall be preserved for three years after the date of
the collision.
(2) A certification that the manufacturer has tested the autonomous
technology on public roads and has complied with the testing
standards, if any, established by the department pursuant to
subdivision (d).
(3) A certification that the manufacturer will maintain a surety
bond, or proof of self-insurance as specified in regulations adopted by
the department pursuant to subdivision (d), in an amount of
$5,000,000.
(d) (1) As soon as practicable, but no later than January 1, 2015,
the department shall adopt regulations setting forth requirements for
the submission of evidence of insurance, surety bond, or self-
insurance required by subdivision (b), and the submission and
approval of an application to operate an autonomous vehicle pursuant
to subdivision (c).
(2) The regulations shall include any testing, equipment, and
performance standards, in addition to those established for purposes
of subdivision (b), that the department concludes are necessary to
ensure the safe operation of autonomous vehicles on public roads,
with or without the presence of a driver inside the vehicle. In
developing these regulations, the department may consult with the
Department of the California Highway Patrol, the Institute of
Transportation Studies at the University of California, or any other
entity identified by the department that has expertise in automotive
technology, automotive safety, and autonomous system design.
(3) The department may establish additional requirements by the
adoption of regulations, which it determines, in consultation with the
Department of the California Highway Patrol, are necessary to ensure
the safe operation of autonomous vehicles on public roads, including,
but not limited to, regulations regarding the aggregate number of
deployments of autonomous vehicles on public roads, special rules
for the registration of autonomous vehicles, new license requirements
for operators of autonomous vehicles, and rules for revocation,
suspension, or denial of any license or any approval issued pursuant
to this division.
(4) The department shall hold public hearings on the adoption of
any regulation applicable to the operation of an autonomous vehicle
without the presence of a driver inside the vehicle.
(e) (1) The department shall approve an application submitted by a
manufacturer pursuant to subdivision (c) if it finds that the applicant
has submitted all information and completed testing necessary to
satisfy the department that the autonomous vehicles are safe to
operate on public roads and the applicant has complied with all
requirements specified in the regulations adopted by the department
pursuant to subdivision (d).
(2) Notwithstanding paragraph (1), if the application seeks approval
for autonomous vehicles capable of operating without the presence of
a driver inside the vehicle, the department may impose additional
requirements it deems necessary to ensure the safe operation of those
vehicles, and may require the presence of a driver in the driver's seat
of the vehicle if it determines, based on its review pursuant to
paragraph (1), that such a requirement is necessary to ensure the safe
operation of those vehicles on public roads. The department shall
notify the Legislature of the receipt of an application from a
manufacturer seeking approval to operate an autonomous vehicle
capable of operating without the presence of a driver inside the
vehicle and approval of the application. Approval of the application
shall be effective no sooner than 180 days after the date the
application is submitted.
(f) Nothing in this division shall limit or expand the existing
authority to operate autonomous vehicles on public roads, until 120
days after the department adopts the regulations required by
paragraph (1) of subdivision (d).
(g) Federal regulations promulgated by the National Highway
Traffic Safety Administration shall supersede the provisions of this
division when found to be in conflict with any other state law or
regulation.
(h) The manufacturer of the autonomous technology installed on a
vehicle shall provide a written disclosure to the purchaser of an
autonomous vehicle that describes what information is collected by
the autonomous technology equipped on the vehicle. The department
may promulgate regulations to assess a fee upon a manufacturer that
submits an application pursuant to subdivision (c) to operate
autonomous vehicles on public roads in an amount necessary to
recover all costs reasonably incurred by the department.