Heck, a mild-mannered Ph.D. in cognitive science with years of expertise in sustainability and transportation, isn’t a reckless driver. But he was demonstrating Nauto, an artificially intelligent tracking device for cars, made by the Palo Alto company of the same name he co-founded 18 months ago.
Like a Fitbit for vehicles, Nauto uses GPS and an accelerometer to track motion, then sends relevant data to the cloud for analysis. But it goes beyond, with cameras facing inward and outward. Nauto, used by car-fleet managers, professional drivers and insurers, evaluates drivers with a credit-rating-like score. At the same time, it uses artificial intelligence and a trove of anonymized data to gain insights into driver behavior. As it learns more, future vehicles using Nauto will better understand the people sharing the road and what they’re likely to do.
“Real human drivers don’t follow the DMV rules to the letter of the law all the time,” Heck said. “If there’s a biker on the right, they’ll cross the yellow median to give him more space. At stop signs, they’ll inch forward to indicate who yields. If the speed limit is 65 but everyone else is going 70, they’ll go 70. They’ll make eye contact with pedestrians and wave.”
The small black device, which looks like an oversize computer mouse, is mounted inside the windshield just above the rear-view mirror. A curvier version with added capabilities will debut in January. The current version, out for just three months, already has thousands of users, Heck said. It costs $399 plus a monthly service fee of $25 to $100.
Toyota, BMW, insurance giant Allianz and another major carmaker last month inked strategic partnerships with Nauto and made undisclosed investments in it. Nauto has $14.6 million in backing.
“Nauto is a great opportunity to better understand driver behavior and dangerous situations,” said Jim Adler, a director at Toyota Research Institute in Palo Alto, in an email. “Autonomous driving will require billions of miles of testing across all types of weather, traffic, and driver conditions (and) trillions of miles through both actual driving and simulation.”
Because Nauto is already in use, Toyota and other companies can gather that data now, he said. Nauto users — all professional drivers for the moment — include Stanford’s Marguerite Shuttle system, San Francisco’s CityWide Taxi, limo fleets and many individual Uber and Lyft drivers, Heck said.
Fleets want to track how their drivers act on the road. Think of it as a high-tech version of those bumper stickers that ask “How’s my driving?”
Alvaro Glenard, owner of Alta Limousine in Palo Alto, uses Nauto on his three limos. While he’d already been using dashcams for several years for safety and liability reasons, Nauto’s real-time monitoring and tracking gives much detailed feedback, he said.
“Any type of erratic driving behavior, such as sudden stops without a collision or hard accelerations, will trigger the alerts I get,” he said. “You don’t want customers to be jerked around within the cars. It’s a step beyond dashcams, which won’t pick up things like that.”
Nauto’s artificial intelligence helps it decide which events are significant enough to warrant uploading video for analysis. Nauto also stores all video for 12 hours so fleets can respond to customer questions or complaints.
Besides reports on driving behavior, Nauto’s next version will issue real-time warnings through sounds and words in dangerous circumstances — if a driver is about to run a red light or hit another car, for instance — but only if its inward-facing cameras determine that the driver is distracted. Artificial intelligence helps it avoid crying wolf, in other words.
“If we see that you are paying attention, we don’t give an alert,” Heck said. “We’re trying to be careful about how often we do real-time intervention.”
Heck sees his company as an evolution beyond Waze, the Google-owned app that gathers alerts from drivers about road conditions. The difference is where a Waze user might report a traffic jam, Nauto can supply video.
“If we treat every car as a sensor and computing device, you could make everyone safer,” Heck said. The next generation of cars will have built-in sensors to collect data which Nauto’s AI can analyze, Heck said.
Artificial intelligence learns from experience, just like people. And most people take seven years to become really good drivers, he said. “We’re building on the real-world experiences of human drivers to optimize autonomous cars.”
Carolyn Said is a San Francisco Chronicle staff writer. Email: csaid@sfchronicle.com Twitter: @csaid
———
©2016 the San Francisco Chronicle
Visit the San Francisco Chronicle at www.sfgate.com
Distributed by Tribune Content Agency, LLC.
