Toyota partners with Stanford, MIT to build self-driving cars

Jonathan Bloom Image
ByJonathan Bloom KGO logo
Saturday, September 5, 2015
Toyota partners with Stanford, MIT to build self-driving cars
Toyota is jumping into the race for autonomous cars with a partnership in the Bay Area with Stanford University and MIT.

EAST PALO ALTO, Calif. (KGO) -- Toyota is jumping into the race for autonomous cars with a partnership in the Bay Area with Stanford University and MIT.

The announcement of the new partnership Friday at the Four Seasons Hotel in East Palo Alto began with a grim statistic.

"A million people die every year in car accidents, one for every year since the invention of stone tools. That's unacceptable," said Toyota's executive technology advisor Gill Pratt at Friday's event.

Pratt used to run the DARPA Robotics Challenge, which makes fierce military robots that do complex tasks with a simple instruction from an operator.

"The situation at Toyota technically is actually quite similar. We have a driver, we have a car, a machine and a person," Pratt said.

And now they have some of the world's best researchers on artificial intelligence.

"What if cars could become trusted partners, partners that could help us navigate a particularly treacherous part of the highway. Watch our backs when we're tired," said Daniela Rus an MIT professor at Friday's event.

Toyota's giving $50 million to Stanford and MIT, to create a center to study the complex problems of a robotic car that has to share the road with humans.

"Work within the human society and human environment is one of the most challenging questions," said Fei-Fei Li, a Stanford computer science professor.

The universities will be working toward a somewhat different vision from the one Google's talked about, rather than cars without steering wheels, they'll be exploring a future where humans remain at the controls.

"I don't think that people will want to completely give up the joy of deciding for themselves exactly how they control the car and drive to where they want to go," Pratt said.

That's also a tougher challenge for researchers. Not just watching the road, but watching the driver, to know when the car should take over to prevent an accident.

"And try to interpret this kind of complex data and learn to predict the patterns and learn to predict the behaviors," Li said.

There are still lots of questions. One audience member asked Pratt how long it will take to predict those behaviors.

"I don't think you'll find a person on earth who has a legitimate prediction for when we'll be done," Pratt said.