MOUNTAIN VIEW, Calif. – In a cavernous garage just off the freeway, a team of young scientists with Stanford University degrees is laboring to create yet another technological revolution.
Drive.ai's hardware and software engineers – six of whom met while doing artificial intelligence research at Stanford -- aim to play a key role in making self-driving cars the first robots humans interact with on a global scale.
“How do you create a robot that people can trust? That’s what we’re working on,” says co-founder and president Carol Reiley, 34, a Johns Hopkins University PhD candidate who paused her studies to move west and wrangle her Stanford-trained peers into startup mode.
This tidy formula of study hall, garage start-up, technological disruption, is well known and beloved in Silicon Valley.
Drive.ai has been working on its tech for a year and comes out of stealth Tuesday with $12 million in funding. It's building advanced software and a kit that can be retrofitted onto standard automobiles to make them self-driving machines, and plans to start testing its amped-up cars soon.
“Our goal is to build the brains of the self-driving car using deep learning to help it think,” Reiley, a former designer of robotic surgical equipment, says.
Self-driving car tech wheeling and dealing
The team is jumping into a market that's buzzing over the prospects for self-driving cars. There's been a flurry of acquisitions as automakers and tech giants race to position themselves.
Ford just announced it will produce a self-driving ride-sharing car with no steering wheel by 2021, and has bought or invested in a variety of small mapping and computer-vision companies.
A few weeks ago, ride-hailing behemoth Uber bought a self-driving truck company Otto, after announcing it would soon be piloting autonomous rides in specially equipped cars from partner Volvo in the Pittsburgh area.
In fact, there are 33 companies large and small working on autonomous car tech, according to CB Insights. They range from Audi to Intel to Nissan to
“In many cases, smaller startups can pioneer new tech faster than big automakers who have that legacy model of building and selling cars,” says Kerry Wu, CB Insights auto tech analyst.
“If your business is facing an existential crisis, spending a billion dollars is a no brainer,” he says.
That was the price
Bleeding-edge tech for self-driving cars will be "the domain of start-ups," said venture capitalist Shahin Farshchi of
"I expect startups to identify the many novel challenges associated with driverless cars, such as low-cost Lidar, sensor fusion (and) computer vision,” he wrote. “Perhaps one company will solve all of these, or maybe several billion-dollar companies will emerge.”
CEO Sameep Tandon, 26, wasn’t yet finished his with Stanford AI studies, but was so taken by the mission that he put his degree on hold.
“I jumped at the chance to get a robot in people’s hands sooner, to be able to see people’s lives improved by AI and robotics,” says Tandon.
'Montessori approach' to teaching cars to think
Leveraging its new Series A funding, Drive.ai is taking a tack similar to billion-dollar Cruise in a self-driving car kit that can be retrofitted onto existing vehicles.
If Reiley and her team have a secret sauce, it would be its advanced software.
Drive.ai is focusing on so-called deep learning algorithms, the similar to technology used by voice-recognition platforms such as Apple’s Siri and Amazon’s Alexa.
“Algorithms (for self-driving cars) used to be purely rules based, ‘If this happens, then you do that,’ but with cars there are millions of cases you can’t plan for,” says Reiley. “With deep learning, you learn by example and previous experience. It’s the Montessori approach to teaching a computer.”
Drive.ai also plans to focus its efforts on an aspect of self-driving cars that isn’t talked about much: how a driverless car communicates with the world around it.
As pedestrians, we receive countless non-verbal cues by looking at the driver of a car stopped at a crosswalk, information that gives us the confidence to proceed. Reiley says her company will integrate a range of audio and visual cues “so everyone understands each other, since we’ll need to coexist.”
Initially, Drive.ai plans to target delivery fleets that cover repeated urban routes, a tactic designed to speed up adoption of its tech.
Over the coming months, it plans to aggressively test its half-dozen tech-laden vehicles around this Silicon Valley town, which is also home to Google’s autonomous cars.
“With different cars operating with different algorithms, it’ll be interesting to see how the social interaction between them happens. Will they be polite and say after you, or cut each other off?” Reiley says with a laugh.
“It’ll be AI versus AI, but ultimately this is a societal problem that needs to be solved," she says. "To safely deploy these vehicles, we need to understand what they’re going to do.”