Laura Bliss is a staff writer at CityLab, covering transportation and the environment. She also authors MapLab, a biweekly newsletter about maps (subscribe here). Her work has appeared in the New York Times, The Atlantic, Los Angeles magazine, and beyond.
The layoffs signal a dramatic scaling back of the company’s autonomous vehicle testing.
More than 100 Uber employees who piloted the company’s nascent self-driving cars in Pittsburgh and San Francisco were laid off on Wednesday afternoon, four months after one such vehicle struck and killed a pedestrian in Arizona.
The layoffs are the latest scale-back of the company’s autonomous vehicle testing program, following other cuts that collectively have slashed the number of autonomous vehicle operators from roughly 400 to 55. Former and current employees expect future testing to follow much more limited routes.
Wednesday’s layoff came during an all-hands meeting at Uber’s Strip District headquarters in Pittsburgh, where the vast majority of the company’s autonomous vehicle operators had been based. The 100-plus affected employees have been given the option of either taking severance or applying for one of 55 more specialized positions involved in AV testing, an Uber spokesperson confirmed.
Vehicle operators in Pittsburgh and San Francisco had been on paid leave since March 18, when an autonomous Uber and its distracted backup driver failed to stop the vehicle from fatally colliding into a homeless woman crossing a Tempe road.
The layoffs came as a surprise to some workers in Pittsburgh, who had expected to hear a solidified timeline for when they would get back to monitoring the wheel. Since March, vehicle operators had been receiving regular email updates from managers informing them that they wouldn’t be needed at work until further notice. But according to affected employees, the general expectation was that they would eventually return, especially after an internal memo in June that outlined a plan for improved testing safety practices indicated a return to the road by August.
“We expected to be retrained,” said one worker affected by the layoff, who declined to be named in order to protect his future employment opportunities.
Others had an inkling. Uber had already dramatically scaled back its testing program, which had employed about 400 operators in about 200 vehicles in five North American cities at its peak earlier this year, following the fatal crash. Arizona blocked Uber from testing on its public roads almost immediately afterward. At the end of March, the company decided not to renew its autonomous vehicle testing permit in California (a spokesperson says it still intends to eventually seek a new permit there, however). At the end of May, Uber officially laid off all backup drivers based in Tempe and Phoenix.
Likewise, testing on public roads in Pittsburgh, San Francisco, and Toronto has been at a standstill since the crash, while the company undergoes internal and external reviews of its testing practices and of the crash itself. Preliminary findings by a National Transportation Safety Board investigation found significant technological errors factored into the collision that killed Elaine Herzberg, a 49-year-old homeless woman who was crossing a multi-lane road by night. But, a subsequent police report found, the crash would have likely been prevented had the backup driver, Rafaela Vasquez, not been distracted. Counter to her job requirements, Vasquez been not looking at the road and had her hands away from the wheel seconds before the crash. Police found that she had been looking down and streaming “The Voice” on a personal cellphone.
Workers affected by the layoff Wednesday now have the opportunity to interview for a newer, more advanced AV testing role, called “mission specialist,” an Uber spokesperson explained. The job will focus more on testing vehicles under far more rudimentary conditions on test tracks and some public roads. Mission specialists will also be tasked with providing more direct feedback to vehicle developers and engineers.
Once testing resumes in August, however, the routes that Uber will be piloting are likely to be significantly scaled back from the expansive “loops” it started on in the fall of 2016. According to two anonymous Uber employees, one current and one former, the company intends to test on public streets in manual mode only, and only a predetermined route between its headquarters and a suburban test track. It will operate vehicles in autonomous mode only on that test track as well as on a short circuit between the two buildings that comprise the Pittsburgh headquarters.
An Uber spokesperson said that the exact routes, and the exact number of vehicles used, will be determined once testing resumes. “Our team remains committed to building safe self-driving technology, and we look forward to returning to public roads in the coming month,” she said.
To improve its testing safety practices in the wake of the March crash, Uber has sought the advice of external experts and consultants; the narrowed focus of the operators who will now man the vehicles may signal a “back to basics” approach that some critics of the old program might applaud.
But one current Uber employee who is unaffected by the layoff said that he’s not so sure the right people got the boot. For months, vehicle operators said they complained to managers about arduous, tedious, and solitary testing conditions that cultivated complacency and exhaustion. Yet the leadership structure at the top of Uber’s Advanced Technology Group remains unchanged. And none of the engineers who might be responsible for the faulty technology that killed Herzberg have been let go, either. “As long as we have the same leaders, we’re going to have the same problems,” he said. ”If we need all this external advice about safety, then we don’t really know about safety.”