Before Google launched its moonshot factory, Uber disrupted the taxi industry, and Tesla delivered its first electric vehicle—Sterling Anderson, Drew Bagnell, and Chris Urmson were working on self-driving cars.
It was a pursuit that many, including some automakers, believed was best confined to science fiction books, not the streets of Pittsburgh or Silicon Valley.
But that’s exactly where self-driving cars—and these early pioneers of the technology—landed.
In the brain trust of self-driving car developers, Anderson, Bagnell, and Urmson are among its most highly regarded. Their positions at Google, Tesla, and Uber, perhaps the most watched autonomous vehicle programs in the world, are an indication of their esteem.
They were working for the big leaguers, helping push self-driving cars out from the fringes of academic research towards a product that people could use in their daily lives.
Now, they’re striking out on their own—and likely competing with at least one of their former employers—with a self-driving car company called Aurora Innovation Inc.
“In a way, Aurora is a coming-back-together story,” said Anderson, who is the company’s chief product officer.
The four-month-old company, which hasn’t revealed its business model until now, is emerging amid a free-for-all environment of billion-dollar acquisitions and investments by automakers and a fight over a limited pool of talent.
It’s the Wild West, and Aurora has already been thrust into the tumult.
Before Aurora could come out of stealth, it was sued by Tesla. The electric automaker alleged Anderson, the former director of company’s semi autonomous Autopilot program, had violated his contract by attempting to recruit at least a dozen Tesla engineers and taking confidential and proprietary information. Anderson has steadfastly denied any wrongdoing.
The lawsuit has since been withdrawn and an agreement reached, leaving Aurora to finally focus on its mission to encourage the broad adoption of self-driving cars.
Aurora wants to provide the full-stack solution for self-driving cars.
Here’s what that means: Aurora will work with an automaker or supplier—”our intent is several,” Urmson told Fortune—to design and develop the right mix of sensors, software, and data services needed to deploy fully autonomous vehicles. Aurora engineers will focus on Level 4, and eventually Level 5, autonomy. Level 4 is a designation by SAE International that means the car takes over all of the driving in certain conditions. In Level 5 autonomy, a vehicle is driverless in any and all circumstances and environments.
The company, which has offices in Palo Alto, Calif. and Pittsburgh, started hiring in January. It’s been a bootstrapped operation so far with $3 million in self-funding and an additional $3.1 million from a group of close advisors. Aurora will launch a Series A funding round in 2017.
Aurora isn’t interested in manufacturing cars; nor does it want to be a volume producer of the many sensors—such as cameras, radar, and light detection and ranging radar known as LiDAR—that these autonomous vehicles need to detect and perceive the world around them.
Instead, the founders see a market need for creating the underlying technology for self-driving cars and helping automakers and others build the services and systems on top of it.
Aurora’s value lies in the design of the sensor and the understanding of what computation needs to be embedded into the main brain for the car, as well as the software and data products.
“This where we see Aurora fitting into the ecosystem,” said Urmson, who is CEO of Aurora. Bagnell, a machine learning expert who will be critical to the Aurora’s mission, is the company’s chief technology officer.
Timing And Competition
This credo puts Aurora in direct competition with Waymo—the Google self-driving project that spun out to become a business under Alphabet. It could also mean going up against Uber as well as competing with dozens of other startups, suppliers, and even automakers that are all working on elements of the grand challenge that is fully autonomous vehicles.
It’s a huge task for any company, even one headed by super stars of the self-driving car world. But Anderson, Bagnell, and Urmson are convinced that they’re not showing up late to the party.
“I think that weighed heavily on all of our minds: Does it make sense to start a company in this space?” Urmson mused. “The short answer is yes. We think the timing is reasonable. Otherwise we wouldn’t have done it.”
“We’re at a pretty unique moment in time in the automotive industry,” Urmson added. “This is not a company I would have started two years ago.”
For now, the company is focused on hiring talent. Aurora is building out a car—a 2017 Audi Q7—that it describes as “pre pre alpha data-gathering platform. The car recently completed a cross-country trip to gather data between Aurora’s two offices. This was a data-gathering journey, not a self-driving one, Anderson stressed.
However, the company does plan on testing self-driving cars on public roads this year.
To understand Aurora’s edge in this space, despite its young age, requires an understanding of its leadership.
The three founders have crossed paths numerous times over the years, although never at the same company. Urmson and Bagnell trace their first meeting to 1999 as graduate students at Carnegie Mellon University, and where they both earned doctorate degrees in robotics. Eight years later, Anderson’s research at MIT would capture Urmson’s interest and respect. Urmson would end up inviting Anderson in 2012 to speak at Google’s self-driving car project.
Urmson, who participated in each of the three autonomous vehicle challenges funded by the Defense Advanced Research Projects Agency, was director of technology of the winning Tartan Racing team in 2007. Bagnell, who Urmson describes as “probably one of the top five applied machine learning people on the planet” was on the same team.
“The three original DARPA autonomous vehicle challenges of 2004, 2005, and 2007 are becoming the Continental Congress of the autonomous vehicle age,” said Reilly Brennan, a general partner at Trucks Venture Capital that invests in transportation startups. “If you were in that oil painting, you are viewed as one of the masters of this era. Because of the embryonic state of the technology at that point, it could be viewed as a proxy for those engineers who were building from a ‘missionary’ perspective, not a ‘mercenary’ one.”
Urmson went on to lead the self-driving car project at Google’s moonshot factory X, a role he held for eight years until leaving in August 2016. Bagnell, an associate professor at Carnegie Mellon (who is currently on leave), helped launch Uber’s efforts in autonomy, ultimately heading the autonomy and perception team at Advanced Technologies Center in Pittsburgh. After finishing a doctorate at MIT and working at McKinsey & Co., Anderson would head over to Tesla, where he would lead the development and launch of the Model X and the automaker’s semi-autonomous Autopilot program.
“It’s powerhouse leadership,”Gartner analyst Mike Ramsey told Fortune. “What Chris knows and discovered in his 8-year journey at Google can’t be replicated. They have a deep level of understanding of what is needed, in terms of integration, and they already know all of the traps.”
Still, Ramsey contends considerable risks remain. “Can they catch up? How much time do they need and how much IP do they actually have?” Ramsey asked.
Window of Opportunity
Despite the risks, the three contend that a window of opportunity has opened in this moment of chaos—a time when major automakers and technology companies are racing to deploy autonomous vehicles, self-driving car startups bursting with talent are popping up every month, and suppliers are trying to find their place in between.
Conflict has started to rise between technology companies and automakers, which have historically owned the vehicle platform, the brand, and the connection to customers, Anderson noted.
A looming question has been who will control access to the customer, the data, ownership of branding for these self-driving cars, especially those used in an on-demand taxi service? The same battle has been playing out with connected car services, as third-party companies—namely Apple and Google—have swooped in to offer in-car software platforms that bring the functionality and feel of a smartphone to the vehicle’s central screen.
For instance, concern over control is what partly drove Toyota to adopt Ford’s SmartDeviceLink technology, an open source version of Ford AppLink instead of Google or Apple’s product. It’s also what prompted a coalition of German carmakers, including Audi, Daimler, and BMW, to acquire Nokia’s mapping unit, HERE.
Aurora is shooting for the sweet spot: it has the talent and experience in solving the technological problems, but doesn’t want or need to control the brand or data. Aurora’s offer could be appealing to automakers even those that already have a partner.
For example, Fiat Chrysler CEO Sergio Marchionne said Wednesday during an earning conference call that the company may seek more partners to help it develop and build self-driving vehicles. Fiat Chrysler already has a partnership with Waymo to rollout self-driving Pacifica Hybrid minivans.
“We need to do this because I think banking all of our solutions on one possible outcome is going to be disastrous,” Marchionne said in reference to its plans to work with companies other than Google on autonomous driving.