On September 20, 2016, the US Department of Transportation (DOT) released its long-anticipated federal policy for the testing and deployment of automated vehicles on public roads. This policy follows and builds upon a guidance document published by the National Highway Transportation Safety Administration (NHTSA) in 2013 (updated in early 2016), which offered recommendations on how state regulators could permit testing of automated vehicles.
The DOT notes that this new policy document is issued “as agency guidance rather than rulemaking,” continuing NHTSA’s strategy of avoiding formal or mandatory regulations on autonomous technologies in favor of establishing a framework for ongoing dialogue between industry, federal and state regulators, the public, and other stakeholders about the right approach in regulating these rapidly developing technologies. As NHTSA Administrator Mark Rosekind remarked to the Autonomous Car Detroit conference in March 2016, “the careful, complex, far-from-rapid processes of formal regulation aren’t necessarily a great match for fast-moving technology advances.” The DOT sees automated vehicles as a potential quantum leap in vehicle safety and has set out in this policy an agenda to embrace and “accelerate the [automated vehicle] revolution.”
While none of the procedures or guidance in the new policy document is mandatory for companies that are manufacturing, designing, testing, or selling automated vehicle systems, the new policy hints at future regulatory actions NHTSA may take and identifies some specific rulemakings that NHTSA expects to initiate in a matter of “months.” The policy is best seen as a stop-gap “to speed the delivery of an initial regulatory framework and best practices” while NHTSA continues to develop more detailed, formal regulations. The new policy states that the DOT will revise the policy document annually based on public comment and its own research efforts.
The policy sets out to accomplish four tasks: (1) provide guidance to industry for establishing best practices for the safe testing and deployment of automated vehicles; (2) offer suggestions for how states should regulate automated vehicles; (3) lay out NHTSA’s current regulatory authorities and toolkit for addressing automated vehicles; and (4) discuss what additional regulatory authorities and tools NHTSA may need to address this new technology.
Companies developing or investing in autonomous vehicle technologies should carefully consider the impact that this new federal policy will have on their products and market plans. First, some aspects of the industry guidance may become mandatory in the future. Second, the Model State Policy contained in the document is likely to influence state regulators. Third, the policy signals NHTSA’s intent to use its existing enforcement authority to regulate automated technologies based on the guidance contained in the policy. And fourth, the policy describes potential federal legislation that could expand NHTSA’s authority in regulating automated vehicles.
The discussion below highlights the key components of the federal policy, which can be found here.
Vehicle Performance Guidance for Automated Vehicles
The first section of the federal policy outlines NHTSA’s Vehicle Performance Guidance for automated vehicles, broken down into 15 discrete safety-related areas. For each area, NHTSA sets the expectation that manufacturers and other developers will collaborate with regulators and industry associations to develop best practices in the design, testing, and validation of automated vehicle technologies, and that they will share data with regulators and each other to help accelerate the development of road-safe automated technologies. To that end, the policy requests that manufacturers and other developers submit a concise “Safety Assessment Letter” to NHTSA for each automated vehicle system that describes the actions they have taken to follow the Guidance in each specific area. The Safety Assessment Letter is currently voluntary, though NHTSA is considering whether to make such reporting mandatory.
As an initial matter, the new policy draws a distinction between “lower-level” automation, in which a human is primarily responsible for monitoring the driving environment, and “highly automated vehicles” (HAVs) in which the automated systems have primary responsibility for monitoring the driving environment. NHTSA has dispensed with its prior 0-4 scale of automation and has adopted SAE International’s standard 0-5 scale, where Level 0 describes a vehicle with no automation and Level 5 describes a completely autonomous vehicle.
The 15 safety assessment areas in the Guidance were drawn up with HAVs (Levels 3-5) in mind and apply only partially to lower-level automated systems (Levels 1-2). However, the policy contains specific guidance for issues facing Level 1-2 vehicles.
The 15 assessment areas are:
- Operational Design Domain (ODD) — The ODD of an HAV describes the parameters in which the HAV is designed to safely operate, such as roadway types, geographic areas, speed ranges, weather and lighting conditions, and other factors. While NHTSA has not yet promulgated standards for testing the performance of HAVs, it expects that an HAV will operate safely in its ODDand that manufacturers will educate human drivers or operators on a vehicle’s ODD in the owner’s manual and elsewhere.
- Object and Event Detection and Response (OEDR) — OEDR refers to an HAV’s recognition and response to various driving events or conditions, such as detecting stopped vehicles in the roadway, responding to road signs or traffic signals, or performing a merger onto a freeway. NHTSA expects that HAVs will be able to respond to the road and traffic conditions that can reasonably be expected within the HAV’s ODD.
- Fall Back (Minimal Risk Condition) — When an HAV encounters a problem—for example, a system malfunction or detection that the vehicle is operating outside its ODD—it should be able to return to a “minimal risk condition,” such as returning control to a human driver or safely pulling the vehicle over to the side of the road.
- Testing and Validation Methods — Manufacturers should develop testing methods through a combination of simulation, test-track driving, or on-road testing that will verify an HAV performs as expected during normal operation and in crash-avoidance or fall back scenarios.
- Data Recording and Sharing — Manufacturers should have a process for documenting events, such as malfunctions or accidents, through black boxes or otherwise, that allows event reconstruction of, at minimum, (1) fatalities or personal injuries and (2) damage causing any motor vehicle to become inoperable. The policy also recommends manufacturers collect data on “positive outcomes” like crash avoidance. NHTSA requests that all manufacturers and other HAV developers share (de-identified) data with each other to facilitate the rapid development of safety features.
- Privacy — Manufacturers should follow best practices, such as the Privacy Principles for Vehicle Technologies and Services published by the Association of Global Automakers, that are designed to protect individuals’ privacy even as manufacturers collect and share data from each vehicle for testing and development purposes.
- System Safety — Manufacturers should implement a robust systems-engineering approach that will build in redundancies to compensate for the potential failure or malfunction of automated systems, especially with respect to software development. Particular design decisions should be based on risk assessments with respect to a vehicle’s “safety-critical system functionality.”
- Vehicle Cybersecurity — Manufacturers should follow industry best practices on cybersecurity and should share information on cybersecurity, including particular vulnerabilities, through the Automotive Information Sharing and Analysis Center “as soon as possible” after such information is discovered.
- Human Machine Interface — Manufacturers should consider how the HAV systems will interact or communicate with any human driver, other operators, or occupants of the HAV, and “external actors” like pedestrians, bicyclists, and other vehicles. With respect to human operators or occupants, the HAV should have some way to communicate that it is functioning, that the autonomous mode is engaged or unavailable at the moment, that the HAV system is experiencing a malfunction, and that the HAV is requesting that a human operator take control of the vehicle.
- Crashworthiness — HAVs should comply with all of NHTSA’s crashworthiness standards, even if the HAV is designed to be unoccupied at all times, such as an autonomous delivery truck that deploys robotic ground drones. NHTSA emphasizes that, even if these vehicles have no people in them, they should comply with, for instance, “geometric and energy absorption crash capability” to protect the occupants of any vehicle that collides with the unoccupied HAV. Manufacturers should also consider developing new safety systems that can use the information provided by advances in the HAV sensing technologies.
- Consumer Education and Training — Manufacturers should educate and train both their own marketing or sales personnel and the consumers themselves on the capabilities and limits of a particular HAV’s technology. NHTSA suggests consumer training might include “hands-on experience,” either on the road or in a virtual reality simulation, to demonstrate the features of the HAV system.
- Registration and Certification — Manufacturers should register, as they already do for other vehicles or components, identifying information about HAVs, and provide on the vehicles themselves “concise information regarding the key capabilities of their HAV system.” If a vehicle’s HAV capabilities are changed or upgraded over time, the registration and on-vehicle information should change accordingly.
- Post-Crash Behavior — Manufacturers should have a process for assessing an HAV for re-deployment after a crash. If sensors or other components of an HAV system are damaged in a collision or other incident, the automated features should be disabled until the vehicle is serviced.
- Federal, State, and Local Laws — Manufacturers should have a plan for how their HAV systems will comply with federal, state, and local laws. In some situations, NHTSA anticipates that an HAV may be programmed to temporarily break a state traffic law by, for instance, crossing double-lines on a roadway in order to avoid a hazard or drive around a disabled vehicle. HAVs should be able to comply with all laws within its operational design domain.
- Ethical Considerations — NHTSA acknowledges that the objectives of safety, mobility, and legality will sometimes conflict and that manufacturers or other developers of HAVs will necessarily program or otherwise influence a decision protocol that implicates ethical considerations. The policy urges these ethical decisions to be made “transparently using input from Federal and State regulators, drivers, passengers, and vulnerable road users.”
As discussed above, this Guidance is directed primarily at HAVs. However, some of the 15 assessment areas apply equally to lower levels of automation (Levels 1-2), and the policy also notes a particular “risk of driver complacency and misuse in Level 2 systems” and the need for “effective countermeasures” designed to prevent that complacency or misuse. Manufacturers should develop testing methods to ensure those countermeasures work as intended. The policy notes that the “technical” distinctions between automation levels may be lost on consumers and emphasizes the importance of communicating the “intended ODD” to operators of Level 2 vehicles—that is, semi-autonomous vehicles that require a human driver to monitor driving conditions.
While the Safety Assessment Letters are voluntary at the moment, NHTSA expects manufacturers and developers who have already deployed HAVs to submit Safety Assessment Letters “within four months” after the Guidance “take[s] effect” following the completion of an approval process pursuant to the Paperwork Reduction Act. And the Safety Assessment Letters may not remain voluntary. One of the “next steps” the policy anticipates NHTSA completing “in the coming months” is “implementing a rule mandating the submission” of Safety Assessment Letters. In the meantime, NHTSA expects manufacturers and other entities to submit these letters for each HAV system that it deploys, and for “any significant update(s)” to an HAV system, including remote software updates.
Model State Policy
The new federal policy introduces its Model State Policy be clarifying the respective scopes of federal and state regulatory authority with respect to HAVs. In particular, the policy “strongly encourages States to allow DOT alone to regulate the performance of HAV technology and vehicles.” The policy notes that, as vehicle equipment increasingly performs “driving” tasks, NHTSA’s authority over vehicle equipment will increasingly encompass task that had previously been regulated under states’ authority to license human drivers. The policy also emphasizes that any state performance standards that conflict with Federal Motor Vehicle Safety Standards (FMVSS) would be preempted by the Vehicle Safety Act. State regulation, the policy stresses, is best restricted to licensing human drivers, enacting traffic laws and regulations, conducting safety inspections, and regulating vehicle insurance and liability.
Apart from the Model State Policy, the federal policy recommends states (1) take an inventory of existing laws and regulations “to address unnecessary impediments to the safe testing, deployment, and operation of HAVs,” and (2) standardize and maintain road infrastructure such as road signs and pavement markings. While the DOT hopes states will adopt the Model State Policy “to avoid a patchwork of inconsistent State laws that could impede innovation and the expeditious and widespread distribution of safety enhancing automated vehicle technologies,” it remains to be seen to what extent states will adopt this Model Policy.
The Model State Policy addresses five areas of state regulatory authority:
- Administrative — The Model Policy recommends states identify a lead agency responsible for the permitting of HAV testing and establish a “jurisdictional automated safety technology committee.” The lead agency should take steps to establish necessary statutory frameworks and regulations relating to HAVs, including a permitting process for testing HAVs in the state.
- Testing — Manufacturers or other developers of HAVs should have to apply for a permit to test HAVs in a state. The Model Policy states that such applications should certify that the HAV “follows the Performance Guidance set forth by NHTSA” and meets applicable FMVSS. In this way, the federal policy suggests how its non-binding guidance, which the policy expressly declares should not be codified by any state, could become mandatory to the extent a manufacturer needs to test its HAVs on a state’s roadways. The Model Policy also recommends that applications for testing should include identifying information for the manufacturer and vehicle, proof of insurance, and a summary of the training of the vehicle’s testers. Applications for testing would be approved by the lead agency and could be granted with conditions. Under the Model Policy, testing would only be performed by designated testing personnel who hold state driver’s licenses and have been trained and instructed about the HAV’s capabilities and limitations.
- Deployment — The Model Policy suggests areas where there may be “gaps in current regulations” with respect to HAVs, including insurance, crash investigation, liability (both civil and criminal), vehicle modifications, and environmental impacts. The Model Policy suggests that title and registration documents reflect that a vehicle is an HAV.
- Law Enforcement — The Model Policy notes that “vehicles that offer less than full automation capabilities” present an increased risk of distracted driving and suggests states promulgate regulations that limit activities such as using electronic devices, eating, drinking, conversing with passengers, and so forth. The Model Policy also notes that first responders should be aware of any risks that might accompany an incident involving an HAV, including “silent operation, self-initiated or remote ignition, high voltage, and unexpected movement.”
- Liability and Insurance — The Model Policy recommends states “consider how to allocate liability among HAV owners, operators, passengers, manufacturers, and others when a crash occurs,” and also which of these entities should be required to carry insurance for the HAV.
Existing Regulatory Powers
The federal policy describes four regulatory enforcement powers that NHTSA can use to address the deployment of HAVs: (1) letters of interpretation, (2) exemptions from existing standards, (3) rulemakings to amend existing standards or create new ones, and (4) enforcement authority to address safety-related defects. In a separate Enforcement Guidance Bulletin issued the same day as the federal policy, NHTSA stated its position that its authority to regulate motor vehicles and their components extends to the software-driven technologies in HAV systems. NHTSA has already signaled in prior publications its intention to expedite the issuance of interpretation letters and exemptions to facilitate the deployment of HAVs, and the federal policy and Enforcement Guidance Bulletin signal a willingness to use NHTSA’s recall authority to address safety issues in automated systems.
- Interpretations — NHTSA can issue letters of interpretation upon request that provide NHTSA’s interpretation of existing statutes or regulations. Letters of interpretation can confirm to the recipient, in the face of ambiguous text, that their actions or anticipated actions are in compliance with the organic statute and regulations. For instance, earlier this year NHTSA interpreted the word “driver” in a particular regulation to include an HAV system.
- Exemptions — NHTSA has the authority to grant a limited number of exemptions from otherwise applicable FMVSS. These exemptions are temporary and limited in number, but could aid manufacturers or other developers in testing HAVs on public roads.
- Rulemaking — Parties can petition NHTSA to initiate a rulemaking for a new vehicle standard, or NHTSA can initiate a rulemaking under its own authority. The federal policy notes that rulemaking is a lengthy process but may be the best option “if a party wishes to avoid compliance with an FMVSS for longer than the allowed time period for exemptions, or for a greater number of vehicles than the allowed number of exemptions.”
- Enforcement — NHTSA has the authority to enforce its vehicle safety standards or to address safety-related defects present in vehicles sold in the United States. As noted in its Enforcement Guidance Bulletin, NHTSA considers HAV technology to fall within the scope of its regulatory authority, and it will “exercise its enforcement authority to the fullest extent” where an automated system causes crashes or has manifested a safety-related defect.
The federal policy includes instructions for how parties can request NHTSA take any of the above actions.
New, “Modern” Regulatory Tools and Authorities
The new policy emphasizes that the Vehicle Safety Act was passed “[f]ifty years ago” and that subsequent technological changes “warrant a review of NHTSA’s regulatory tools and authorities.” Accordingly, the policy lists potential changes to its statutory authority that would enhance its ability to regulate HAVs. While the policy emphasizes that the “DOT does not intend to advocate or oppose any of the tools discussed” in the document and is only seeking public comment, these additional authorities could form the basis for new legislation that increases NHTSA’s authority over vehicle manufacturers.
The potential new authorities identified in the policy include: (1) pre-market reporting by vehicle manufacturers; (2) pre-market approval authority; (3) cease-and-desist authority; (4) expanded exemption authority (i.e., a greater number and duration of exemptions); and (5) post-sale authority to regulate software changes. In addition to those new authorities, NHTSA suggests several potential tools it could be provided with legislative amendments, including: (1) variable test procedures allowing for the testing of HAVs in unpredictable testing environments; (2) the power to require modifications to vehicle designs “as necessary to reduce high-level risks to acceptable levels”; (3) additional recordkeeping and reporting requirements; and (4) enhanced data collection tools, such as the requirement of advanced event data recorders. Finally, the federal policy notes challenges NHTSA has faced with its limited resources, noting in particular NHTSA’s limited ability to retain outside experts and hire qualified in-house experts.
As discussed above, the DOT intends this federal policy to be an evolving document to be updated on an annual basis. The policy also outlines some of the next steps the agency will undertake to further the deployment and regulation of HAVs, including:
- The agency will continue to receive comments about the policy through a public notice-and-comment process and through public workshops and expert reviews.
- The agency will also continue to perform research and gather data related to automated vehicles, and it will develop a mechanism by which various parties can share anonymized data in HAV testing and development.
- With regard to the Safety Assessment Letters, the DOT will complete an approval process under the Paperwork Reduction Act, publish a template letter for manufacturers and developers, and may implement a rule making such letters mandatory.
- The DOT will consider other possible rulemakings, including ones that would (1) establish an objective method for determining an HAV’s level of automation under the SAE International scale; (2) require registration of and reporting from HAV testers and operators, and (3) update or establish new FMVSS that would permit the certification of vehicles that have no human-operated controls.
- With regard to the Model State Policy, the DOT will continue to work with state regulators and internationally to facilitate the adoption of consistent regulations.
- NHTSA will also solicit public comment on its proposals for new federal regulatory authorities to address HAVs.