To test passenger reactions, the ride-hailing company will put a handful of self-driving cars into service in San Francisco starting Wednesday, even as the parent company of industry pioneer Google inched closer to commercializing its technology on Tuesday by spinning out a stand-alone business unit devoted to self-driving vehicles.
Uber has spent three months testing self-driving cars with passengers in Pittsburgh, using hybrid Ford Fusions and plug-in hybrid Volvo XC90s, a pilot that will continue. Both there and in San Francisco, an Uber engineer sits in the driver’s seat, ready to take the wheel or hit the brakes, while another engineer monitors what the vehicle “sees” on a laptop. Uber couldn’t say when the cars might operate without backup drivers — something not currently legal on California roads.
Uber’s apps will randomly assign self-driving cars to San Francisco passengers who will pay standard rates and can opt for a regular car instead. With more than 30,000 Uber drivers trawling the Bay Area (albeit not all at the same time), the chances of getting one of the “fewer than 10” self-driving cars may be akin to winning a golden ticket to Willy Wonka’s chocolate factory.
“Expanding our self-driving pilot to San Francisco allows us to continue to improve our technology through real-world operations,” wrote Anthony Levandowski, head of Uber’s Advanced Technology Group, in a blog post. “San Francisco comes with its own nuances, more bikes on the road, high traffic density and narrow lanes.”
As for the California DMV's reaction to the news, it encourages the responsible exploration of self-driving cars, according to a statement released to media outlets on Dec. 13. "We have a permitting process in place to ensure public safety as this technology is being tested," the statement reads. "Twenty manufacturers have already obtained permits to test hundreds of cars on California roads. Uber shall do the same."
And on Dec. 14, the Associated Press reported that California regulators told Uber to stop the self-driving service until it gets the required permit.
Uber’s autonomous Ford Fusions have been cruising San Francisco streets for weeks to create detailed 3-D maps of the cityin preparation for the pilot.
On a South of Market test drive on Tuesday, a retrofitted Volvo seemed to perform well, swerving around complicated construction barriers and making lane changes — a recent feature, Levandowski said, along with handling rain and recognizing cyclists in bike lanes. In one possible glitch that the Uber engineers termed “slightly aggressive,” the car accelerated as it approached a yellow light, causing the backup driver to hit the brakes.
Overall, the ride was noticeably smoother than a September test drive in Pittsburgh in the earlier-generation Ford Fusions. During that ride, Uber engineers took control of the car every couple of minutes. With the more recent Volvos, an engineer took brief control about three or four times over 20 minutes.
“The ride is supposed to be boring; if it’s exciting, something went wrong,” said Brad Templeton, a Silicon Valley entrepreneur who was an early strategy and engineering consultant on Google’s self-driving efforts. “It can be underwhelming.”
He sees the Uber pilot and Google spinoff as incremental but positive steps forward. “Having real customers and seeing how they interact is very important,” he said.
In contrast to Uber and other rivals, Google’s parent company has offered no timeline for commercialization of its self-driving technology. But its move to create a self-driving division — dubbed Waymo, short for a way forward in mobility — suggests that the company is pushing forward more forcefully.
“We are close to bringing this to a lot of people,” said Waymo CEO John Krafcik at its media event on Tuesday at Google’s offices in San Francisco. He ticked off possible uses such as ride-hailing, trucking, personal ownership and public transportation.
Waymo’s head of technology, Dmitri Dolgov, declined to comment on a Bloomberg report Tuesday that Waymo and Fiat Chrysler will offer a ride-hailing service with semi-autonomous minivans as early as late 2017.
Many people remain skeptical of self-driving technology, even as Silicon Valley and Detroit move toward implementing it. Of about 1,600 Americans surveyed by PwC this year, roughly half said they believe “autonomous cars are dangerous,” and 28 percent think the cars could be hacked.
Uber is trying to gradually accustom people to self-driving vehicles, just as it did for ride-hailing, said Douglas Schmidt, a professor of computer science at Vanderbilt University. “At first people were reluctant to use Uber, so it came up with ways to try to get them hooked on it,” he said. “They’re doing the same thing now to get the driverless car model mainstream, getting early word of mouth to spread.”
The self-driving Volvos in Uber’s pilots stem from a $300 million partnership between it and the Swedish car maker.
“We have 90 years of experience with automotives,” said Marten Levenstam, Volvo’s vice president of product planning, at a media event Tuesday. “We know how to design a car. You can clearly see how well-executed this self-driving Uber is; it is far from a science project. It’s a real prototype.”
Uber’s self-driving efforts took off in early 2015 when it poached dozens of robotics experts from Pittsburgh’s Carnegie Mellon University. By contrast, Google started working on driverless cars in 2009, and its bubble-shaped cars have logged more than 2 million miles. But Google has seen defections by key engineers and executives frustrated that the technology is taking so long to mature.
Levandowski, who co-founded and led Google’s self-driving car team, is a case in point. He left in January to co-found San Francisco startup Otto, which develops kits to make trucks self-driving. Otto quickly got autonomous big-rigs on test drives in Nevada and Colorado, and was acquired by Uber this summer for a reported $680 million, a deal that put Levandowski in charge of Uber’s self-driving efforts.
Now Waymo, with Google’s technology, appears to be stepping on the gas. By spinning off, it will be able to reward employees with lucrative incentives, similar to those offered by startups, said Krafcik, the CEO.
Waymo is focused on self-driving technology and is not a car company, he said. “We are not in the business of making better cars,” Krafcik said. “We are in the business of making better drivers.”
Schmidt said that approach mimics how Google has handled smartphones, focusing on Android software rather than hardware. However, he noted, eventually Google did start making its own phones, “so I wonder whether they’ll stay out of the car market in perpetuity.”
Waymo executives emphasize that their technology is fully autonomous. However, Krafcik said Waymo cars may still have a steering wheel and other controls, due to regulatory issues.
Waymo’s mapping system not only identifies items and people on the road; it can also predict their future actions, the company said. For example, a Waymo car can identify a police car and “see” whether its lights are flashing. It can pick out a cyclist and predict what direction that cyclist will go next. The self-driving car is prepared for unexpected encounters, like a truck driving on the wrong side of the road or an animal darting in front of it. And it’s learning to cope with all kinds of weather conditions, including heavy rain and snow.
The company said it reached a milestone in October 2015 when a blind man rode by himself on a public road in a Google self-driving car without a steering wheel, accelerator or brakes in Austin, Texas.
“I want very much to become a member of the driving public again, and this technology is going to bring that to pass,” said that man, Steve Mahan, former CEO of the Santa Clara Valley Blind Center, at Waymo’s event on Tuesday. “So I’m excited to be alive now and to have this exposure to the technology. I am looking forward to the distribution of these vehicles into the driving arena.”
©2016 the San Francisco Chronicle Distributed by Tribune Content Agency, LLC.
