Hit with fatal crash case, Uber's vision of autonomous cars begins to blur

After Dara Khosrowshahi took over as Uber’s chief executive last August, he considered shutting the company’s money-losing autonomous vehicle division. A visit to Pittsburgh this spring changed that.

In town for a leadership summit, Mr. Khosrowshahi and other Uber executives were briefed on the state of the company’s self-driving vehicle research, which is based in Pittsburgh. The group was impressed by the progress its autonomous division had made in testing driverless cars in Pittsburgh and in Arizona, according to three people familiar with the ride-hailing company, who were not authorized to speak publicly. They left the meeting energized, convinced that Uber needed to forge ahead with self-driving cars, the people said.

But days after the summit, one of Uber’s autonomous cars struck and killed a woman who was pushing a bicycle across a street in Tempe, Ariz. Video from the March 18 collision showed a distracted safety driver failing to react in time as the vehicle barreled into the pedestrian, Elaine Herzberg.

The accident threw Uber’s autonomous vehicle efforts into flux, immediately forcing the suspension of its self-driving car tests in cities including Tempe, Pittsburgh and Toronto. Months later, Uber’s executives are divided over what to do with the autonomous business, according to the people familiar with the company. While one camp is pushing Mr. Khosrowshahi to seek partnerships or even a potential sale of the unit, known as the Advanced Technologies Group, a rival contingent is arguing that developing self-driving technology is crucial to Uber’s future, the people said.

Mr. Khosrowshahi remains undecided, the people said, though he has expressed a desire to partner with other companies on autonomous technologies. In recent months, Uber has started talking with a few auto manufacturers about potential partnerships, including supplying Uber’s autonomous driving technology for use in Toyota’s minivans, according to one person familiar with the talks. Toyota declined to comment.

The internal debates are unfolding at a time when many companies can ill afford to pause on autonomous technology given stiff competition from carmakers and other tech companies. In recent months, top engineers have left Uber’s self-driving project for lucrative opportunities elsewhere. Uber’s self-driving cars recently returned to the road in Pittsburgh but with human drivers at the wheel, meaning employees are driving around like any other motorist — except their vehicles are carrying hundreds of thousands of dollars in technology.

The issue of whether to retain or sell A.T.G. is complicated by Uber’s stated intention to go public by the end of 2019. The company, valued at $62 billion, has racked up billions of dollars in losses since it was founded in 2009 and needs to persuade investors that it can eventually create a sustainably profitable business. The self-driving efforts, which have been losing $100 million to $200 million a quarter, do little to help that case. And Mr. Khosrowshahi has been shedding money-losing businesses since he joined Uber.

At a meeting in Pittsburgh on Aug. 8, according to a person briefed on the event, Mr. Khosrowshahi did not address what he would do with the self-driving efforts but told employees there that it “is a big-time hardware manufacturing, software problem at scale. Lots of tech companies out there are going after this problem, but I think there are very few companies who are taking this on end-to-end at scale the way we are.”

In a statement, Uber said: “Right now the entire team is focused on safely and responsibly returning to the road in autonomous mode. That’s our No. 1 objective, and we have every confidence in the work they are doing to get us there.”

Uber first made its interest in self-driving cars public when it hired about 40 researchers and scientists from the National Robotics Engineering Center at Carnegie Mellon University in 2015. At the time, the company’s chief executive was one of the founders of Uber, Travis Kalanick, who had decided to bet big on self-driving vehicles. He wanted to prepare Uber for a future when fleets of driverless cars could move passengers efficiently and safely around the clock.

In 2016, Uber acquired Otto, a self-driving truck start-up whose founders had decamped from Google. The deal later spurred a trade-secrets-theft lawsuit from Google’s onetime self-driving car unit, Waymo. The case briefly went to trial this year, generating headlines and embarrassing revelations, before Uber settled with Waymo in February.

In its rush to get on the road with driverless cars, Uber also ran afoul of regulators. The company started testing its autonomous vehicles in San Francisco in 2016, without a permit from California’s Division of Motor Vehicles. The state agency ordered Uber to apply for a permit, but the company refused, saying permits were not necessary since safety drivers were monitoring the cars. The D.M.V. ultimately revoked the registrations for the 16 self-driving cars that Uber was testing in the city.

By early this year, Uber’s self-driving division was preparing to ramp up development, pushing its testing cars in Arizona to tally more miles. The goal, according to internal documents reviewed by The New York Times, was for Uber to win regulatory approval to start testing a self-driving car service in Arizona before the end of this year.

But the crash in March — the first known fatality involving a pedestrian and an autonomous car — altered everything. Since then, Uber has steadily narrowed the scope of its autonomous vehicle operations.

In May, Uber announced that it was shutting its driverless testing hub in Arizona and laying off 300 employees. A day later, preliminary findings from federal regulators investigating the crash confirmed what many self-driving car experts suspected: Uber’s self-driving car should have detected a pedestrian with enough time to stop, but it failed to do so. Uber has begun a safety review and plans to publish its assessment in the coming months.

Mr. Khosrowshahi has started to subtly de-emphasize the company’s role in developing driverless technology.

At a conference last year, he said it was a “huge advantage” for Uber to have its own autonomous technology while operating a global ride-sharing network. But this May, Mr. Khosrowshahi said that while Uber needed to have access to autonomous technology, it aimed to be “neutral.” He said Uber would be open to licensing its own technology or building around alternatives from other companies — a stark contrast to the company’s previous approach of owning and operating the entire self-driving “stack” of technology and hardware.

And in July, Uber announced that it was closing its autonomous trucking business. The company instead said it would focus exclusively on building self-driving cars.

“For now, we need the focus of one team, with one clear objective,” Eric Meyhofer, who leads Uber’s driverless car efforts, wrote in an email to employees.

In the preceding months, some senior engineers and executives with expertise in self-driving vehicles had already left. One of those was Don Burnette, one of Otto’s founders, who became the chief executive of a new self-driving company called Kodiak, which focuses on long-haul trucking.

“I really wanted to focus on the trucking problem, and there was not as much focus on that at Uber,” Mr. Burnette said.

He added that Uber would most likely continue to pursue its vision of driverless cars because it and other companies “have been working on it for so long, promising this for so long, and they have a tremendous amount of money behind them.”