Automakers didn’t build the self-driving car: Google did. That’s a big problem for them. Hoping to catch up, Ford, Toyota, and Volkswagen are betting on academics. Along with Nvidia, Samsung, Qualcomm and Panasonic, they’re each giving $300,000 to the University of California at Berkeley to fund artificial intelligence research.
The alliance, called DeepDrive, is a rare moment of AI cooperation among car companies, which are racing one another to create the kind of brains that propel Google’s prototype gumdrop-shaped vehicles around Mountain View. It also highlights the new position universities find themselves in. Their AI lab work is in high demand—and corporations don’t want to wait months or years to get their hands on it.
The companies’ money will go to projects selected by UC-Berkeley. In return, the automakers get to give feedback on research proposals; meet the academics toiling away on the tech; and, thanks to the upfront payment, can commercialize any of the research without having to go through the headache of an additional licensing stage.
“They’ve essentially pre-negotiated access to software,” said Trevor Darrell, a professor at the university who leads DeepDrive. In corporate terms, $300,000 may not seem like a lot of money, but altogether the donations will back between 20 and 30 graduate students a year.
It’s a cheap way for the companies to get a bead on a dangerous, unpredictable future. “If vehicle manufacturers, five years from now, haven’t been to the drawing boards to figure out how to get self-driving tech into their cars, then those companies will be left out,” said Thilo Koslowski, top car analyst at the research firm Gartner. “It’s that dramatic.”
For UC-Berkeley, it’s an opportunity to get funding without having to delay publication at the behest of a sponsor. That’s the normal protocol for corporate-backed research, where companies can ask to review results prior to general publication, delaying publication for months– or more.
Openness has become a big deal in artificial intelligence as the pace of research speeds up. No one wants to be caught reinventing the wheel [or the car]. With DeepDrive, the university “can have open research with no publication restrictions, no lock down of early review for patenting, so the research can move as fast as it possibly can,” says Darrell.
It also gives the university a way to test its theoretical ideas in the real world, says Pieter Abbeel, a professor at UC-Berkeley and one of the principal investigators at DeepDrive. “We do all this research, but unless you do a startup where’s it going to go?” Abbeel said. “These companies, they all have experts in the same topics but might not have the time to do research. But they understand everything we’re doing and can translate it very nicely into their own projects and we can see it in action.”
Through the project, UC-Berkeley researchers could also get access to driving data from the companies, and be able to run their software on the automakers’ vehicle simulators, letting them test out new approaches without risking crashing real cars, he said.
The types of problems DeepDrive’s researchers will tackle read like an index page from a science fiction novel: custom semiconductors for vision systems, software to predict how a pedestrian will behave, AI that can drive in unusual terrain, techniques that let machines learn from human drivers. It will also fund work on Caffe, a programming framework developed by UC-Berkeley that cuts out some of the grunt work in writing AI code.
DeepDrive is emblematic of the new interest in university research projects around artificial intelligence and robotics, says Andrew Moore, the dean of Carnegie Mellon University’s school of computer science and a former Google employee. “We’ve hit this inversion where the stuff the universities are doing is applicable right now to industry,” he said.
Typically, university research needs to be adapted for specific industries or problem areas. That’s not the case with the current crop of AI technology, which is generating such excitement because it can be applied easily and quickly to new areas. “It’s a real boom time,” Moore said. Google and Facebook regularly turn research papers into products in a matter of months, as opposed to years. In 2014, Uber partnered with Carnegie Mellon to develop self-driving car technology. By 2015 it had hired away academics from the lab.
Car companies have been throwing a lot of money and resources at artificial intelligence lately. Ford moved the development of self-driving cars from its research lab to its engineering operations in June, as the company prepared to put AI into more of its vehicles. In September, Toyota hired Gill Pratt, the U.S. military’s top robotics engineer, as part of a $1 billion investment into developing smarter cars. Volkswagen AG used its presentation at the Geneva auto show in February to talk about its planned investments in self-driving cars.
The automakers are starting to worry that if they let Silicon Valley win the AI race, they’ll be sidelined as contract manufacturers of sheet metal– a future they’d like to avoid. So, as with past research pushes like hydrogen fuel cells, the competitors are coming together.
“You need to think differently about who is the partner and who is the enemy,” says Gartner’s Koslowski. “The whole notion of frenemies will become more important to the auto industry going forward.”
Most of all, it gives them a better chance to take some of the marketing sheen away from Google, which has garnered huge amounts of media attention for being the first company to create and demonstrate an autonomous vehicle. That’s likely galling to the auto companies which have been developing self-driving, or driver-assistive services (think cruise control) for years, but haven’t been so public with their zanier research efforts. After all, the team that won a government self-driving challenge in 2007 was sponsored by General Motors, Caterpillar, and Continental. (Then again, the director of that project, Chris Urmson, now runs Google’s self-driving car group.)
“Google has had a lot to do with this in terms of their ability to get out and really publicly talk about what they’re doing,” says Tom West, the director of the California Program for Advanced Transportation Technology at Berkeley. “I think a lot of the auto companies were doing a lot of this work and were doing it under wraps.”