Lyft is quickly amassing partners for its autonomous car effort. The latest one is Drive.ai, a company born out of Stanford’s artificial intelligence lab that uses deep learning to turn ordinary vehicles into self-driving vehicles.
Drive.ai recently announced it had raised $50 million and added Google Brain founder and former head of Baidu’s artificial intelligence division, Andrew Ng, to its board. So far, Drive.ai has six autonomous cars outfitted with autonomous driving kits, but the company declined to say how advanced its technology is, only that it’s working toward getting it to capably drive in all environments, no humans necessary.
Drive.ai has had cars on California roadways for the last year, accruing miles and data to better train its algorithms. What it doesn’t have is a ready supply of people to which they can roll its technology out—that’s where Lyft comes in.
Lyft offers companies a way to slowly expand autonomous driving to the masses “in a safe and reliable way,” says Drive.ai cofounder Carol Reiley. The test pilot with Lyft will take place in the Bay Area and focus on training Drive’s cars to coexist with pedestrians, bicyclists, buses, and all the other minutiae involved in navigating urban landscapes. The main goal, says Reiley, “is to figure out how autonomous vehicles are going to impact the public, and how they will adopt autonomous driving.”
Though lots of companies are investing in building cars that drive themselves, few seem concerned with how consumers will accustom to hailing, getting into, and sitting idle in the back of a robo-chauffeured car. Drive.ai is trying to be conscious of this. Part of its kit is a digital billboard of sorts that sits atop the car and communicates with passersby. The company has tried using emoji-base signaling, to avoid language issues, and “safety sounds,” to allow drivers to send different messages to pedestrians, from angry honks to courteous acknowledgements.
Taggart Matthiesen, head of product at Lyft, argues that the Lyft platform will also make it easier to determine the optimal moment to introduce a customer to an autonomous car, based on information about weather, road conditions, time of day and other data. “We have access to all these cities, and we can evaluate a number of environmental constraints prior to dispatching the vehicle, and so that allows us to slowly but surely open this up.”
Though the ride-hailing company began its foray into self-driving research much later than its largest competitor, Uber, and trails behind it in terms of spending, it has quickly racked up partnerships with car manufacturers and other self-driving players like Waymo, GM, nuTonomy, and Jaguar Land Rover. (Uber’s self-driving project, meanwhile, has partnered with companies like Daimler and Volvo.)
Lyft’s approach to self-driving is in some ways very analog. The company sees its future in liaising between human riders and robot drivers, and that means learning how to adapt the tech to riders’ needs. “It’s not just about us plugging into other people’s technology—we’re not looking to test this on passengers,” says Matthiesen. “Our goal is to provide a higher level of services.”