The robot cars are coming. But will they let me drive?
Nissan’s head of autonomous vehicle research, Maarten Sierhuis, said the question of who takes the wheel, and when, will be determined by circumstance. On a routine morning commute to work, the robot car might handle the tedium of the drive. But on a scenic spin along Highway 1, the human might take the wheel.
“It doesn’t have to be all the time,” Sierhuis said in remarks Wednesday at the Code/Mobile conference at The Ritz-Carlton in Half Moon Bay, Calif. “This is what we’re doing at Nissan. Between the person driving versus autonomous there is this incredible space … a sliding autonomy.”
Nissan’s idea is in stark contrast to that of Google’s, whose prototype gondola-shaped autonomous car had no steering wheel.
Sierhuis and Qualcomm’s vice president of strategic development, Chris Borroni-Bird, talked about the challenges that confront carmakers as they seek to adapt to a changing world, and the threat of industry-wide disruption that the entry of tech players such as Google and possibly Apple represent.
“It is going to happen. They will have to change,” said Sierhuis of the traditional automakers. “The question is how long will it take for them to need to change?”
Sierhuis, an expert in the field of artificial intelligence, leads Nissan’s research into autonomous driving and connected vehicles. Before joining the Japanese automaker, his career in research and software engineering took him to NASA, Xerox Palo Alto Research Center, NYNEX Science & Technology and IBM. Earlier this year, Nissan and NASA announced that they would work together to develop autonomous vehicles that could tool across town — and, potentially, the surface of Mars.
Borroni-Bird worked as director of advanced technology vehicle concepts for General Motors, helping Chrysler build a prototype of a car whose on-board fuel processor could convert gasoline to hydrogen, carbon dioxide and water. He also played a role in the electric-powered EN-V, a two-seat concept car that is both small and emissions free, which was showcased at the Auto China 2012 show. At the San Diego chipmaker, he’s responsible for Qualcomm Halo, which is exploring new ways to wirelessly charge electric vehicles through technology embedded in parking spots and roadways.
Facets of the autonomous vehicle are already creeping into vehicles, in the form of proximity sensors that detect objects in the motorist’s blind spot, automatic braking the vehicle or parking it. Researchers like Sierhuis are grappling with new problems, such as how a car’s sensors could be alerted if a pedestrian is about to step out into traffic, or how cars could communicate with one another to navigate a four-way stop.
Often, the human element is the hardest for an autonomous car to anticipate.
“How do you predict what a human is going to do — that is a hard thing,” said Sierhuis. “That’s where the AI planning comes in. It’s not just planning what the car is going to do, it’s taking into consideration what everyone else is going to do.”
The robot cars will have to contend with impossibly confounding scenarios like the one Creative Strategies’ analyst Tim Bajarin posed, where an autonomous vehicle approaches an intersection with a pedestrian, a bus full of children and failing brakes.
“It’s a very good question. A lot of people are still wrestling with how to answer that,” said Borroni-Bird.
This article originally appeared on Recode.net.