Uber AV crash spotlights federal inaction on safety

“Would you be ready for your organization if it were involved in a tragic, high-profile, fatal crash?”

uber crash

Volvo XC90 SUV involved in crash. (Photo Credit: Uber)

A wide-ranging critique of state and federal AV safety standards and an Uber (NYSE: UBER) mea culpa shaped a panel discussion on lessons learned from a high-profile 2018 crash involving one of the tech giant’s vehicles.

Ensar Becic, a project manager with the National Transportation Safety Board (NTSB), said that as investigators examined the roles of developers, states and the federal government in the Tempe, Arizona crash they kept returning to one question:

“How to determine the adequacy of risk and mitigation strategies since there are no standards and assessment protocols.”

A devastating experience


Becic spoke during a panel discussion at this week’s Automated Vehicle Symposium. Also participating was Uber’s Advanced Technology Group (ATG) director of organizational safety, Christopher SanGiovanni, who discussed the corporate overhaul that occurred in the wake of the crash, in which an Uber test car, a Volvo XC90 SUV, killed a pedestrian.

“I want to kick off with a rhetorical question for developers,” SanGiovanni said. “Would you be ready for your organization if it were involved in a tragic, high-profile, fatal crash? It’s unimaginable. You can’t underestimate how devastating it is to those at the organization.” 

The Uber crash is widely viewed as a turning point for the industry, as companies started to shift their rhetoric from getting a self-driving vehicle on the road as quickly as possible to touting the importance of safety and an abundance of caution as they put self-driving cars and trucks on public roads.

Lax federal oversight


Weak federal oversight remains a huge problem, Beck said. He decried the lack of even a basic test for autonomous vehicles, “be it an obstacle course, a perceptual test or tangible requirements such as testing for miles or adherence to development standards.”

The National Highway Transportation Safety Administration’s (NHTSA) automated vehicle guidance contains 12 relevant safety elements for testing, and NHTSA encourages AV companies to submit safety self-assessments based on these areas.  

But that guidance does not contain metrics for autonomous driving system (ADS)  developers to determine whether they achieved safety goals tied to those 12 elements, Becic said. What is more, submission of the report is voluntary, and NHTSA does not evaluate or approve those reports.

Of the 66 companies that have received permits to test vehicles in California alone, only 23 have submitted safety self-assessment reports showing how they operate their vehicles.

“Some are quite detailed and technical,” Becic said, “while some are marketing brochures.”

Making safety assessments mandatory

Many of Becic’s discussion points appeared in an NTSB report last year on the Uber crash. Since then the U.S. Department of Transportation has released new federal guidance on autonomous vehicles, as well as an AV transparency initiative, neither of which contains language about tightening safety standards.

NTSB, an independent federal agency, has recommended that NHTSA mandate the safety self- assessments, and evaluate and approve those reports prior to testing so states can decide whether to allow such testing.


Some states have started to develop AV legislation, but although many have policies that promote AVs, few have requirements for testing. 

“We view such approval as providing a minimal safeguard for testing,” Becic said.

A tech giant overhauls its safety culture

From an industry perspective, companies appear far more willing than they were in the past to embrace safety standards and regulation as a prerequisite to AV adoption and public acceptance.

In a separate panel discussion this week, Gretchen Effgen, vice president of Go-to-Market and Marketing for  Hyundai-Aptiv, said the key is balancing regulatory oversight with innovation, with the former “not going so far as to dictate an outcome.”

Uber has implemented numerous technical and operational safety changes since the crash —  such as allowing the anti-braking system on its self-driving Volvos to remain active (the system was disabled at the time of the crash) and adding a second operator to each vehicle, SanGiovanni said during the panel discussion.

But “it was important not to just look at the obvious,” he said. “You need to dig deep, ask the right questions.”

Following recommendations from NTSB, the company implemented an aviation-style safety management system, a top-down organizational approach to risk management. The organizational process goes beyond engineering, SanGiovanni said, to look “at the risk of the entire enterprise we are putting on the road.”

Technology components are part of a self-driving vehicle, he explained, which “executes behaviors on the test track and road.” But that vehicle itself is part of an even broader “self-driving enterprise” that includes everything from operator effectiveness to organizational impacts on the enterprise, said SanGiovanni. 

Probing your organization’s safety culture, its underlying assumptions and values, SanGiovanni cautioned, is “really difficult.”

On the road again, with plenty of speed bumps

After temporarily shutting down operations in the wake of the crash, Uber’s self-driving cars are on the road again in Pittsburgh, Dallas, Washington and Toronto.

But the “challenge of vehicle automation has not been solved,” Becic said, and ADS testing on public roads still has inherent risk. Noting that the Tempe investigation was not necessarily about Uber ATG, he said the report and Uber’s willingness to share its experience will hopefully lead “to a positive impact.”

Related stories: 

Ike files voluntary safety report revealing details of autonomous truck operation

Infrastructure is next big thing in AV investment

Feds announce new AV initiative to improve transparency

Click here for more articles by Linda Baker.

Exit mobile version