Most Americans aren’t ready to trust self-driving vehicles. If that’s going to change as automated travel nears, consumer advocates say that federal regulators responsible for protecting the motoring public must set and enforce rigorous safety standards for the fledgling technology. But there are growing concerns the government is doing little to ensure autonomous vehicles are ready for the road—and that recent revisions to the Federal Automated Vehicles Policy have undercut the National Highway Traffic Safety Administration’s role as a safety watchdog.
Representatives from prominent consumer groups expressed those worries last week during a meeting with federal transportation officials in Washington, D.C., where government officials are already working on an update to the two-month-old revised policy. Further weakening of any federal oversight, the consumer groups warn, will backfire in the long run.
“Whether it’s because of General Motors ignition switches, Takata airbags, or Volkswagen emissions software, consumers are not necessarily going to immediately trust auto companies when it comes to something as fundamental as handing over the driving task,” said William Wallace, policy analyst at Consumers Union. “Consumers are not necessarily going to assume what companies are saying about the safety of automated driving systems is true. They’re going to want proof.”
Instead of setting safety benchmarks, legal policies, and standards that would compel such proof, in the form of vetted performance data and detailed information on automated systems, the Department of Transportation’s revised automated-vehicles policy requires no such thing. It merely (meekly?) asks automakers and tech companies to submit a voluntary self-assessment of their efforts to develop and enhance safe vehicles.
This Navya driverless shuttle bus was involved in a minor collision within 45 minutes of its launch last week in Las Vegas.
It was first introduced by the previous administration in September 2016, and the policy’s voluntary character has continued under Transportation Secretary Elaine Chao (pictured above). Her revised policy, Automated Driving Systems: A Vision for Safety 2.0, contains a couple of noteworthy differences. The original requested that companies explain how they consider safety in relation to 15 key areas of development, including data recording, system safety, and ethical considerations; that has been whittled to 12 areas. Also, the revised policy makes it abundantly clear that companies are required to do absolutely nothing. It emphasizes that “no compliance requirement or enforcement mechanism” exists and that it should be “clear that assessments are not subject to federal approval.”
The Transportation Department’s efforts on autonomous vehicles are particularly important right now, because legislation proposed in the U.S. Senate would consolidate the federal government’s power to regulate their safety and would largely prohibit state and local governments from setting their own benchmarks.
“Transparency builds trust. No company should be afraid of transparency if they’re putting safety first.”
– William Wallace, Consumers Union
“To our great disappointment, the second iteration of the automated-vehicles policy is nothing more than voluntary guidance that the industry may completely ignore,” said Peter Kurdock, director of regulatory affairs at Advocates for Highway and Auto Safety, an alliance of consumer groups, public-health professionals, and insurance industries that seeks to reduce traffic deaths. “In fact, the agency clearly states that the voluntary guidance has no compliance or enforcement mechanism.”
So far, only one company—Waymo—has submitted a safety assessment. At 43 pages, Waymo’s document offers overviews on topics such as how its self-driving system detects and responds to obstacles in the road, development of its testing and validation, and information on how vehicles transition into fallback mode following an unexpected incident. Kurdock said the document amounts to little more than a marketing effort that doesn’t include substantive information.
Waymo announced last week that it will remove human safety drivers from some of its test vehicles operating on public roads in the Phoenix area.
Beyond the safety assessments themselves, the broader importance of data in the self-driving realm gained prominence in September when the National Transportation Safety Board released its final report on a fatal crash involving a Tesla Model S sedan that operated with its Autopilot feature engaged. That particular feature is an advanced driver-assist system and not a driverless one, but the recommendations that stemmed from the investigation nonetheless hold implications for fully driverless cars.
In the final NTSB report, the agency recommended that the Department of Transportation define the data parameters needed to probe automated vehicle-control systems involved in a crash. Further, the agency recommended that NHTSA use those parameters to ensure that all new vehicles equipped with automated systems can capture that data. At a minimum, the NTSB report said, data should be provided to its own investigators and NHTSA regulators. Wallace argues it should be available to everyone.
“Proponents refer to its potential in almost mythical terms,
as if the introduction of [autonomous] vehicles will magically make 37,000 yearly deaths disappear overnight.”
– Jason Levine, Center for Auto Safety
“Transparency builds trust,” he said. “No company should be afraid of transparency if they’re putting safety first. The competitive push should not overwhelm the importance of transparency and cooperation on safety. That will come back to bite the industry.”
A dearth of information could ultimately hinder consumer acceptance of driverless technology. Already, surveys show that Americans are reluctant to embrace autonomy. In March, 75 percent of respondents told AAA they’d be afraid to ride in a self-driving car. More than half say they’d be leery even of sharing the road with an automated car while driving a traditional vehicle themselves. In a separate study, 55 percent of survey respondents told global technology consulting firm Gartner they would not consider riding in a fully autonomous vehicle, citing worries about technology failures and cybersecurity.
With autonomous features already reaching the road and deployment of fully autonomous vehicles planned within the next three years, whatever action the federal government now takes—or does not take—will hold great sway over how the technology rolls onto public roads and into the hands of consumers.
“Proponents refer to its potential in almost mythical terms, as if the introduction of these vehicles will magically make 37,000 yearly deaths disappear overnight, but the public, however, is incredibly skeptical,” Jason Levine, executive director of the Center for Auto Safety, told the NHTSA panel. “There may never be a more critical time in autonomous-driving technology in terms of consumer acceptance.”