It is important to distinguish between those features that enhance safety versus those that provide safety and convenience is how Princeton’s Dr. Alain Kornhauser set the stage for the Smart Driving Car Summit session, Incentivizing Through Regulation. Regulations must ensure that the comfort features, such as Automatic Cruise Control (ACC), do not degrade safety. He also believes that consumers should not have to “read the fine print” to understand the Operational Design Domain (ODD) limitations of a given feature.
Building on Kornhauser’s premise, Bryant Walker Smith, Associate Professor, University of South Carolina School of Law & School of Engineering, suggests the measurement of success should be viewed in the number of new problems created versus the problems solved. As straightforward as that sounds, unintended consequences, both positive and negative, are often difficult to envision.
For instance, Walker Smith points to the horse externality problem (PDF) of the early twentieth century and how the internal combustion engine was the “green” answer that solved the mess on the streets of NYC and reduced the food demands of the equine workers.
And regulation to affect a particular outcome is tricky, as illustrated by Walker Smith’s example of a mail carrier who circumvented the law of a Massachusetts city that had banned cars in the early 1900s. The mail carrier simply hooked up his horse to pull his vehicle through the town when he traveled to that ODD.
Testing, Testing & More Testing #
California was one of the early states with autonomous vehicle legislation with the passage of SB 1298 in September 2012. Bernard Soriano, Deputy Director, California DMV, is responsible for overseeing the Golden State’s autonomous vehicles program, which was a result of that legislation. The data the DMV collects as part of that program provides a glimpse into how much each company is testing and how the autonomous driving systems evolve over time.
The AV testing program has seen 285 mostly minor collisions from 902 active vehicles that have driven 8.54M miles driven since 2016. The encouraging trend is that the number of disengagement events has been decreasing on a per-mile basis over time. Waymo, with one disengagement per 29,944 miles, had the fewest engagements per mile in 2020.
Soriano credits a mindful and collaborative process, that featured extensive outreach to the public, governments (e.g., NHSTA), and to industry, for the program rules. This approach seemed to quell the concerns when the legislation was passed that it would drive autonomous vehicle companies out of California. As of March 2021, there are 56 manufacturers approved to test autonomous vehicles with a safety driver present, 6 of those are approved to test without a safety driver and one (Nuro) is approved for commercial deployment of its autonomous delivery vehicles.
The LA Times’ Russ Mitchell brought up the recently released emails between Tesla and the California DMV regarding Tesla’s beta release of FSD City Streets. Although Soriano was not authorized to comment on the Tesla/DMV email exchange, the emails clearly state that Tesla’s testing is for Level 2 capabilities, which are outside the scope of the DMV’s test program. Tesla says that they did not perform autonomous vehicle testing in California in 2020.
“For Reporting Year 2020, Tesla did not test any vehicles on public roads in California in autonomous mode or operate any AVs, as defined by California law.”
Further, all those expecting a Tesla Robo-chauffeur will have to keep waiting, as Tesla admits in its 12/28/20 correspondence that:
“Autopilot is an optional suite of driver-assistance features that are representative of SAE Level 2 automation (SAE L2).”
This points to a bigger issue of when and how to communicate to the public and what is the process for doing so, says Bryant. Soriano stresses that the learnings from real-road testing must be presented transparently and in a form that the public can consume.
Alison Pascale, Senior Policy Strategist, Audi of America says that informing and engaging the public about the future of transportation is the focus of PAVE. Clearing the Confusion (PDF) is an example of PAVE’s effort to provide a common language for autonomous features. PAVE is a coalition of industry organizations, non-profits, and academics with 65+ members. Notably absent from this group is Tesla.
Federal versus State Regulation – Where to Draw the Line? #
Mobility as a Service and Transport as a Service will be the initial use-cases for autonomy (driverless) and not personal ownership is the view of Pascale. Although she terms the U.S. as AV-friendly, she says that states are leading the federal government in creating rules and testing requirements.[1] Pascale speaks from experience, as prior to her role at Audi, Pascale was Director of Governmental Affairs, Policy and Strategic Planning at the National Highway Traffic Safety Administration (NHTSA).
She points out that each state is unique and, as a result, policies differ based on geography. This is not necessarily bad as Compass Technologies’ Richard Mudge suggests that one of the secrets of the economic success of the United States is that it allows for 50 different experiments in adopting new technologies and processes. He points out that autonomous vehicles are still in a nascent stage and that the federal role in the regulation of automobiles was late in their deployment (e.g., the 1960s for vehicle safety requirement).
What is the demarcation between state and federal regulatory authority as autonomy transforms transportation, ponders Walker Smith? Traditionally, the federal government (NHSTA) was responsible for the design, safety, and performance of the vehicle, while the states set rules around its operation (e.g., insurance, liability, traffic laws, registration, etc.). Will the regulatory distinction between federal and state authority be as clear-cut when there are fleet management systems driving purpose-built vehicles in mobility as a service-type application?
Three Steps for the Feds #
Although she thinks it is important to have federal vehicle standards, instead of state-by-state, Pascale agrees that it is still too early for specific federal regulations. She says that industry groups (such as SAE, IEEE, and ISO) need to coalesce around standards first. Pascale has three recommendations for the federal government to speed the rollout of driverless vehicles[2]:
- Echoing the Reason Foundation’s Marc Scribner, she recommends increasing the FMVSS (Federal Motor Vehicle Safety Standards) exemptions for driverless and novel designs. The current U.S. Code sets a 5,000 vehicle limit that NHTSA may exempt during a two-year period time.
- Harmonize federal and state policies.
- Accelerate time to complete rulemakings. She points out that NHSTA could provide information on the data that they are looking for from vehicle manufacturers and provide timelines for what to expect, which would help speed the decision process.
Pascale argues that purpose-built, driverless vehicles potentially offer positive societal benefits (e.g., accessibility, low-cost mobility, and goods delivery, etc.) and are worthy of exemptions. Citing the 1.5 years it took Nuro to receive an exemption for its R2X goods delivery robot, she says the exemption process is too long and uncertain.
Pascale does not seem to hold out much hope for near-term legislation, since it was stalled in the last Congress. Without new legislation, NHSTA could still provide exemptions for large pilot programs with data sharing requirements. She touts the benefits of such an approach are that it would help prove out business models, allow more innovative developments while providing a larger sample of test data to understand how the vehicles work in the real world.
She believes that NHSTA has the resources to support such exemptions if they make it a priority. Soriano emphasized that resources will always be in short supply, particularly from the technology perspective. Building off the idea of limited resources, Diana Furchtgott-Roth, Adjunct Professor of Economics, George Washington University emphasizes that governments should set performance standards and not picking technology solutions.
Spectrum for Safety #
With its November 2020 adoption of new rules for the 5.9 GHz band, it could be argued that the FCC is picking a technology solution for the deployment of Intelligent Transportation System (ITS) services. In the FCC’s words from its press release (PDF),
“The new rules also will improve automotive safety by reserving the upper 30 megahertz of the band for Intelligent Transportation System (ITS) services and designating C-V2X as the technology standard for safety-related transportation and vehicular communications… Today’s action therefore begins the transition away from DSRC [Dedicated Short-Range Communications] services—which are incompatible with C-V2X—to hasten the actual deployment of ITS services that will improve automotive safety.”
As background, in 1999 the FCC set-aside 5.85-5.925 GHz for ITS, or as the USDOT calls it the 5.9 GHz Safety Band. DSRC (IEEE 802.11p/SAE J2735) set the standard for V2X communications (e.g., vehicle-to-vehicle, vehicle-to-infrastructure, vehicle to pedestrians/cyclists). C-V2X was the creation of the 5GAA, which was formed from a group of vehicle and telecom equipment makers. Cellular-V2X (C-V2X) as initially defined as LTE V2X in 3GPP Release 14, operates on a device-to-device basis like DSRC, but adds connectivity to existing cellular networks and by extension, cloud services (see the 5GAA website for a concise explanation of the technology).
As expressed in her recent City Journal column, Furtchgott-Roth echoes the concerns of many that the FCC’s November ruling may harm the rollout of safety applications by reallocating 45 MHz of the band for unlicensed use (i.e., Wi-Fi). The concern goes beyond the potential for limiting the usefulness of the band, but it extends to the potential for harmful interference from the adjacent Wi-Fi devices.
The FCC’s rulemaking is supposed to be entered into the federal registry at any time. Furtchgott-Roth points out that there is a 30-day clock for opposing parties to appeal the rule once it hits the registry. Furtchgott-Roth is probably as close to this rulemaking as anyone as she was the Deputy Assistant Secretary for Research and Technology at the US Department of Transportation. She says it is unprecedented for the entire Executive Branch to oppose the FCC’s actions, as they did with the FCC’s plan for 5.9 GHz.
Similarly, the Departments of Commerce, Defense, and Transportation were aligned against the FCC’s (PDF) unanimous, April 2020 decision to allow Ligado Networks to deploy a low-power terrestrial nationwide network in the 1526-1680 GHz (L Band) that GPS also uses:
“Second, the public interest would be furthered best if the Commission evaluates this new harmful interference metric under more scientific rigor before applying it to a large number of GPS devices relied upon for national security and public safety, including civil GPS receivers owned and operated by emergency first responders and others for a variety of critical functions.”
Their concerns were echoed by the aviation, manufacturing, and various companies in separate filings with the FCC. On January 19th of this year, just before the change in administration, the FCC denied the request for a stay from the NTIA.
Ligado Networks points to the bigger challenge of GPS resiliency, which has been another area of focus for Furtchgott-Roth. Just before leaving the USDOT in January, her team released a 457-page report that summarized a multi-year, multi-agency, multi-stakeholder report (PDF) on its testing of 11 different technologies to complement existing GPS to provide redundancy and resilience in the event of spoofing and jamming.
Like the previous sessions, there were enough unexplored threads in this session to warrant additional panels. In the meantime, stay tuned for next week’s Smart Driving Car Summit session, where experts will examine Human-centered Design of Safe and Affordable Driverless Mobility.
[1] NHSTA has its AV Test Initiative which is a clearinghouse of sorts that allows states and companies to voluntarily submit information about their automated vehicles and testing. https://www.nhtsa.gov/automated-vehicles-safety/av-test-initiative-tracking-tool
[2] The Alliance for Automotive Innovation provides a four-year action plan with 14 specific recommendations on its website. https://www.autosinnovate.org/avroadmap
*The word”fully” is a nod to Bryant Walker Smith’s humourous introductory remarks about how phrases in the autonomous vehicle world are routinely modified by that adverb.
