Vehicle Automation and Fully Self-Driving Cars

Background

Cars today are incorporating increasing levels of automation to enable a car’s computer—rather than the human driver—to accomplish key driving tasks. For example, consumers can now buy partially-automated cars that can park themselves, keep themselves centered in their lane, automatically brake to avoid a fender-bender, and speed up and slow down to keep pace with traffic, as long as a driver still behind a wheel is able to take over as needed. As the technology evolves, cars will be able to complete all driving tasks with human supervision. And eventually, they will be fully self-driving, meaning that they will drive themselves without human oversight.

Fully self-driving cars are also known as autonomous vehicles (AVs). SAE International, formerly known as the Society for Automotive Engineers, developed the following classification system that the National Highway Traffic Safety Administration uses to describe the different levels of automation:

  • Level 0—no automation: The driver is responsible for all aspects of driving.
  • Level 1—driver assistance: One aspect of driving is automated, with the driver still in control of the car. For example, a vehicle might be able to control speed or keep itself from moving into another lane, but not both.
  • Level 2—partial automation: More than one aspect of driving is automated, with the driver still in control of the car. For example, the car could control both steering and speed.
  • Level 3—conditional automation: The car takes over driving, including by controlling speed, steering, and braking. But a human driver must be ready to take control quickly.
  • Level 4—high automation: The car takes over driving without the need for human intervention. But it can do so only under certain conditions, such as in sunny weather.
  • Level 5—full automation: The car takes over all aspects of driving in all conditions. Human driving is completely eliminated.

If policymakers act with intention, this technology could fundamentally and positively alter society. But these outcomes are not guaranteed.

Among the potential benefits are:

  • Expanded mobility for underserved populations: Fully self-driving cars may increase mobility options for people who do not drive today, such as people with disabilities, people with low incomes, and people living in rural areas. Expanding access will require policymakers, manufacturers, and key stakeholders to work proactively. For example, without concerted attention, fully self-driving cars may not incorporate universal design or perform appropriately on unpaved or unmarked roads in rural areas. In addition, policymakers need to determine how to ensure that people with low incomes have access to the benefits of fully self-driving cars despite their expected high retail price. Without planning, an influx of fully self-driving cars could decrease the ridership of public transportation, undermining what in many areas is a prime source of affordable transportation. 
  • Increased safety for all: Automated vehicle technology continues to evolve to provide safety benefits, such as automatic braking to avoid crashes and fender-benders. As we evolve toward fully self-driving cars, these benefits may increase. For example, fully self-driving cars can’t be distracted or drunk. But neither are they perfect. They can also make mistakes, so crashes will not be eliminated. Ensuring safety—both for those in the car and pedestrians, cyclists, and others outside the car—will be most challenging during the transition period in which roads will have to accommodate different levels of automation.
  • Improved livability: If policymakers implement appropriate land use and pricing policies, fully self-driving cars may lead to communities with less traffic congestion, more room for bike lanes, improved safety for pedestrians, denser developments, and more green space. Some of these benefits are due to these cars’ sensors, which allow fully self-driving cars to drive much closer to other cars and in narrower lanes. Fewer parking lots and garages may be needed, particularly if fully self-driving cars spark more shared use of cars—and especially if multiple people ride in a car at once. Instead of cars being idle all day in parking lots, they could be used continually to ferry people to and from their destinations.

On the other hand, fully self-driving cars have the potential to decrease livability by contributing to sprawl. For example, people might not mind long car rides because they can use that time productively instead of driving. Fully self-driving cars could also undermine the infrastructure for other forms of transportation. For example, shared use might lower the cost of each car ride so much that people choose to take an autonomous taxi rather than walk, bike, or take public transportation. As a consequence, traffic congestion may increase.

Thus, some of the benefits of fully self-driving cars will only be realized if people share vehicles and continue to use high-capacity public transportation in dense urban areas. Some industry leaders expect that car ownership will greatly decrease once fully self-driving cars are deployed and that the shared use model will dominate.

Achieving these potential benefits will require the engagement of federal, state, and local policymakers.

Federal policymakers are responsible for ensuring vehicle safety. As such, they hold primary responsibility for setting up flexible yet enforceable standards that will allow the industry to continue to innovate while ensuring vehicle safety and consumer protection. This has not happened to date. For example, existing regulations continue to require steering wheels and brake pedals, which will not be necessary with fully autonomous vehicles. A key challenge has been a lack of technical knowledge among regulators. To combat this, some have suggested that policymakers, automakers, consumer groups, and other stakeholders work in partnership to create dynamic and flexible safety standards that promote both innovation and safety.

Today, ensuring driver safety is the responsibility of state policymakers, who do so primarily through licensing. In the future—when cars are drivers, in the case of fully self-driving vehicles—this may mean that states play a role in ensuring vehicle safety, particularly if the federal government does not do so effectively.

Both state and local policymakers work to ensure road safety through such measures as setting and enforcing appropriate speed limits and designing streets to protect pedestrians and cyclists. They also play an important role in pricing the transportation system, such as by setting tolls (see also Financing Livable Communities section of this chapter.)

Moving forward, state and local policymakers will continue to make important infrastructure decisions that will have a large impact on automated vehicle technology. For example, connected infrastructure that communicates with automated vehicles can promote safety and create less congested roads. Policymakers will also need to consider how to properly price the transportation system in order to promote societal benefits of the technology and avoid its potential pitfalls.

A key issue that remains is how to establish consumer trust in automated driving technologies that will form the pathway to fully self-driving cars. Surveys show that many Americans appear reluctant to give up the control they have as drivers. This presents challenges in establishing the trust that will be needed to deploy fully self-driving cars in sufficient numbers to realize all the potential benefits. Consumer education and training are essential. Fortunately, automatic features slowly are being incorporated into new cars. By integrating education and training as these features come out, manufacturers can encourage users of all ages to learn gradually how to use these technologies, develop trust, and adopt new driver behaviors.

As cars incorporate more automated features, consumer issues take on increased importance. Among them are privacy, security, and liability in case of crashes.

Automated vehicles collect a large array of data on passengers and the surrounding environment. They already are becoming more integrated with other connected devices riders use, such as cell phones. This provides them access to even more data. Companies could generate income by selling access to this information. Consumers need to understand what information is collected, how it may be used or sold, and what their rights are to stop companies from collecting, using, or selling their data.

Data security issues develop in part because automated vehicles may transmit and receive information from other cars and infrastructure, such as traffic lights. Protecting the system from security hacks with the potential for great harm is essential.

Although automated and fully self-driving cars are expected to be safer, crashes and injuries that require a determination of liability will still occur. As automated cars (and the hardware and software comprising them) continue to take on more responsibility for driving, determining liability in the case of a crash becomes more complex. This is particularly the case with partially automated vehicles (in which the driver must over driving when needed). In such cases, the human driver or the manufacturers (of the car itself or the hardware or software comprising it) could all be partially or fully liable for a crash. With fully self-driving cars, in which the car is fully responsible for driving, there is recognition that the manufacturers will be held responsible.

A related question is how the insurance marketplace will evolve. As the cars themselves become drivers, passengers may need to carry insurance, or the vehicle manufacturer or owner may do so.

VEHICLE AUTOMATION AND FULLY SELF-DRIVING CARS: Policy

Access to and safety of fully self-driving cars

In this policy: FederalStateLocal

Policymakers should work in partnership with a diverse range of stakeholders, including consumer groups and the private sector, to ensure that automated vehicles achieve their promise, including:             

  • Expanding mobility options for all, particularly currently-underserved populations. This includes older adults, people with disabilities or limited mobility, people with low incomes, and rural residents.
  • Increasing safety for all road users. This includes pedestrians and cyclists. Increasing safety should be achieved through rigorous testing and careful public deployment.
  • Expanding and providing continuous consumer education. This will empower consumers to take advantage of the full benefits of automated technology as it develops and will ensure its effective and safe use.
  • Embedding consumer protections to safeguard the best interest of consumers. This includes areas such as access to redress, privacy, and security.
  • Enhancing communities to make them more livable, walkable, and age-friendly. Transportation infrastructure that is no longer needed should be converted to new uses that enhance the ability of residents of all ages and all ability levels to participate in community life.

Expanding mobility options for all, particularly currently-underserved populations:

Policymakers must ensure that autonomous vehicle deployment promotes a more equitable transportation system that improves mobility for all. This includes shared mobility options (see also Mobility on Demand—Transportation Network Companies policy).

Policymakers should use tools such as incentives, subsidies, and mandates to expand access for underserved populations (see also Financing Livable Communities section of this chapter).

AVs should complement, rather than compete with, existing public transportation infrastructure (see also Mobility on Demand—Universal Mobility as a Service policy in this chapter).

In order to ensure access for people with disabilities specifically, the following policies apply:

  • Policymakers, the private sector, and people with disabilities must collaborate to ensure that accessibility is fully integrated into the design of autonomous vehicles and system deployment.
  • Commercially-produced autonomous vehicles should ensure accessibility for people with disabilities and those with limited mobility by incorporating universal design into the product development and delivery process.
  • Companies providing AV transportation services should provide wheelchair-accessible services on a basis that is comparable with services provided to people who do not have such a need, as required under the Americans with Disabilities Act.
  • Policymakers should consider how to make human assistance available and affordable for those using AV transportation services who need it.
  • Autonomous vehicles should achieve a predictable and acceptable level of performance and safety so that a driver’s license is not required to operate vehicles defined as operating at SAE levels 4 and 5.

Increasing safety for all road users:

Policymakers and the private sector must work together to increase safety for all road users, including pedestrians, cyclists, and others outside the vehicle. The private sector should continue to innovate technologically, with policymakers providing sufficient guardrails both to allow the innovation to flourish and to protect consumers. Regulation should be dynamic and flexible to respond to changing needs as the technology evolves.

Federal policymakers should set outcome-oriented performance goals to achieve key autonomous vehicle safety targets, such as progressively decreasing fatalities and serious injuries from crashes. The private sector should innovate to meet these targets. If the federal government fails to ensure automated vehicle safety, such as by failing to set appropriate measures and goals, state and local governments should have the ability to do so.

In addition, state and local governments should ensure the safety of those inside and outside the vehicle through traffic laws, road information, communications technology infrastructure, road design and infrastructure, engineering, and transportation planning. (see also the Effective Planning, as well as the Financing Livable Communities sections of this chapter). 

Policymakers should study how best to harmonize the federal, state, and local efforts, as well as to identify optimal testing that should be done and by whom.

AV regulations should be coordinated among federal, state, and local governments to increase efficiency and effectiveness.

Deployment should be limited to the conditions and speeds that have been shown to be safe through rigorous publicly-reported testing, with results shared with regulators in a format that allows them to verify the data independently. Ideally, testing should be standardized. Crashworthiness standards should be updated and enforced to ensure safety.

Partially automated vehicles (SAE levels 1-3) should incorporate driver monitoring technology to ensure that a driver is prepared to take over as needed without distraction. They should also integrate technology to make it easier for drivers to take control when needed, including providing sufficient advance warning. Federal policymakers should establish minimum performance goals for driver engagement.

Crashes must be reported to regulators as soon as possible. Federal regulators should be able to take immediate action, without having to engage in administrative procedures, to require manufacturers to fix major vehicle safety risks that could cause serious injury or death, including by stopping sales of vehicles with such risks. This is known as having “imminent hazard authority.”

Regulators should investigate crashes and have the authority to suspend testing programs to ensure safety.

Expanding and providing continuous consumer education:

Consumers must be empowered with ongoing education and training (beyond simply disclosing information in the owner’s manual). Consumers should understand the capabilities and limitations of the technology, as well as the safety standards that are applied to the vehicle. Manufacturers should provide consumer education at point of purchase and delivery, with continuing driver education as needed, particularly when automatic software downloads enable new features.

Embedding consumer protections to safeguard the best interest of consumers:

Policymakers and the private sector should embed consumer protections into automated vehicles and related services (see also Chapter 11, Financial Services and Consumer Products - Consumer Principles, as well as this chapter’s section on Mobility on Demand—Transportation Network Companies policy).

Policymakers and the private sector must safeguard access to redress after injury or damage, including through insurance and access to the courts. As the transition toward autonomous technology evolves, policymakers should consider whether new insurance or liability models are needed (see also Chapter 11, Financial Services and Consumer Products - Insurance, Chapter 12, Personal and Legal Rights - Private Enforcement of Legal Rights, and Chapter 12, Personal and Legal Rights - Pre-Dispute Mandatory Binding Arbitration).

Policymakers and the private sector must ensure the digital security of automated vehicles. They must make all possible efforts to protect them from harmful digital attacks, unauthorized access, and damage. Policymakers should require a comprehensive and layered system for preventing, detecting, and responding to cybersecurity threats. Companies should adopt the National Institute of Standards and Technology’s Cybersecurity Framework or a similar industry-standard cybersecurity framework (see also Chapter 11, Financial Services and Consumer Products - Security of Connected Devices policy).

Policymakers and the private sector must ensure privacy protections. Consumers should not be required to share any non-essential personal information in order to use AVs or related services. Consumers should maintain control over their individualized personal information that is not already publicly available. They should be able to understand how that information is collected and decide how it may be used (see also Chapter 11, Financial Services and Consumer Products - Digital Privacy and Security). However, policymakers and the private sector should be able to use de-identified aggregated data for research, evaluation, and planning purposes.

In order to maximize consumer choice and increase affordability, policymakers should refrain from requiring manufacturers to use a specific engine type, such as by limiting deployment to electric vehicles (see also Chapter 10, Utilities—Ratemaking).

Enhancing communities to make them more livable, walkable, and age-friendly:

As AVs are deployed, policymakers should seize the opportunity to redesign local communities to enhance livability for all. AV developers should share de-identified data with policymakers to aid in this planning.

Policymakers should convert infrastructure that is no longer needed, such as parking, to new community uses. New uses might include mixed-use developments, parks, and community centers.

In making infrastructure investment decisions, policymakers should consider what is needed for AVs and those outside the vehicle, such as pedestrians and cyclists.

Both the benefits and burdens of new infrastructure should be equitably distributed across all communities (see also Chapter 10, Utilities—Ratemaking).

Policymakers should promote shared use of autonomous vehicles (see also the Financing Livable Communities section in this chapter.)