Aurora will develop a self-driving system that can work with cars from a number of companies.
(TNS) -- Few startups start out with the pedigree of Aurora Innovation.
The self-driving car company’s founders are considered among the best engineers in their fast-growing field.
Chris Urmson used to lead Google’s groundbreaking autonomous vehicle project, now known as Waymo. Sterling Anderson shepherded Tesla’s Autopilot system — which can steer cars at freeway speed — to its release in 2015. Machine-learning expert Drew Bagnell was one of the key engineers in Uber’s self-driving effort.
They quietly formed Aurora in December and at first kept details of their venture under wraps. Tidbits of information, however, emerged when Tesla sued Aurora in January, accusing Anderson of stealing confidential data and encouraging other Tesla engineers to jump ship. Anderson called the suit meritless and said he stole nothing.
That suit is behind them, after Tesla and Aurora reached a settlement last month with no admission of wrongdoing. The startup is now opening up about its plans.
Aurora will develop a self-driving system that can work with cars from a number of companies, Urmson and Anderson say. The focus will be on software rather than hardware, although Aurora may work with partner firms to improve the sensors that autonomous vehicles need to see the world around them.
The idea is to get self-driving cars on the road quickly, working in partnership with traditional automakers rather than designing and building cars from scratch. The young company has already had discussions with some automakers, Urmson said.
“Having a technology is not particularly interesting if it’s not out there being used,” he said, in an interview at the company’s spartan Palo Alto headquarters. “We think we have some of the world’s best people and have some very good ideas about how to solve the driving part of it. We’d love to marry that up with people who really understand what it takes to build and ship vehicles.”
Unlike some in the tech industry who consider the major automakers too slow, Urmson has great respect for car companies and their ability to build good, complicated products at a massive scale.
“People talk about Silicon Valley and complexity and innovation,” he said. “Well, these guys are churning out this thing that comes off the line once every two minutes and runs for 15 years.”
As the former head of Tesla’s Autopilot program, Anderson already has experience developing a technology, integrating it into cars and getting it into drivers’ hands. But he argues that he can have a wider impact with Aurora than with his former employer.
“It was clear that a lot of the rest of the industry was not coming along as quickly as I would have liked,” Anderson said. “So this was about finding a way to accelerate that.”
Aurora joins a crowded field of companies — global automakers and small startups alike — trying to perfect autonomous vehicles. And yet, while most of the public attention has focused on a few prominent names, such as Google and Uber, no one truly dominates the market, said Douglas Schmidt, associate chairman of computer science and engineering at Vanderbilt University.
“It’s not like you’re trying to go in there and knock off an incumbent with an installed base,” he said. In addition, venture capitalists are eager to invest money in the field, Schmidt said.
“This is a time to strike out on your own, because there’s a lot of potential upside and not much downside,” he said.
Aurora has about $6 million in private funding and plans to raise its first round of venture backing soon, Urmson said. The company occupies a modest space next door to the General Motors Advanced Technology Silicon Valley Office off El Camino Real. (That’s a coincidence, Urmson said. It was the only available building Aurora could find in the area that had both office space and a garage.)
More than most fields, autonomous vehicle research in the last decade has revolved around a relatively small clutch of talented engineers whose paths have often crossed. That’s certainly true for Anderson, Urmson and Bagnell.
All three earned doctorates in robotics from powerhouse engineering schools: Carnegie Mellon University for Bagnell and Urmson, Massachusetts Institute of Technology for Anderson. Bagnell and Urmson collaborated on Carnegie Mellon’s team for the 2007 DARPA Urban Challenge, a robot-car race sponsored by a wing of the U.S. military. Urmson thought highly enough of Anderson’s work that he had him address the Google self-driving team while Anderson was still at MIT.
Although Urmson counts his time at Google as “the best seven years of my life,” he decided last year that it was time for something new. He quit Google in August and spent several months talking to venture capitalists and others, trying to figure out his next step. When he decided to form a self-driving startup, Anderson was a logical partner.
“I had a lot of respect for the work that was going on at Tesla, and I thought, hey, if I was going to go start something, and it was going to be in self-driving cars, wouldn’t it be great to have someone who seems like a great person and has experience actually launching stuff?” Urmson said.
“There’s only a small group of people who have really carried this stuff for a long time, before it got very trendy and VC money was being thrown at it,” he said.
Aurora has not yet started testing its system on public roads, but it has equipped a few Audi Q7s with sensors to gather data. The company does not want to get too tied to any one car or brand, however.
“It’s important to us, for our own development, that the platform we’re building not be over-fit to one particular make or model,” Anderson said.
While they want some flexibility in the system, Anderson and Urmson say their ultimate product will probably require a standard set of sensors and computing power. It will be up to automakers to integrate that into their vehicles.
The race to develop self-driving cars has placed traditional automakers in an odd position.
Many have in-house research programs working on autonomous vehicles. At the same time, they are forging partnerships with outside firms — many in the Bay Area — developing their own versions of the technology.
And while each company, large and small, has its own software twists, all are writing programs to do essentially the same jobs — accurately perceive the world around the car and, based on that information, decide what to do next.
So will each automaker eventually offer its own version of the same basic technology? Or will one system prove so superior that all others will fall away, with every car company leasing it?
“We expect that there will be a handful of solutions, but probably not more than that,” Urmson said.
©2017 the San Francisco Chronicle Distributed by Tribune Content Agency, LLC.