The Japanese automaker hires “all-star group of scientists and engineers”
Toyota Research Institute CEO Dr. Gill Pratt shed more light on Toyota’s high-tech, U.S.-based think-tank, as well as the experts working there.
Founded two months ago with a $50 billion investment, TRI is billed as a bridge between research and development, has two U.S. offices — one near Stanford, the other MIT — and already 30 research projects are underway. Pratt, speaking at the Consumer Electronics Show in Las Vegas, revealed the nature of two of these projects, both focused on autonomous vehicles and the challenges autonomous-vehicle development faces.
“Although the industry has made incredible strides over the last five years,” he said, “we remain a long way from the finish line of truly autonomous cars. Most of it has been relatively easy, because most driving is easy. Where we need help with autonomy is when it’s hard.”
One TRI autonomous research project is dubbed “Uncertainty on Uncertainty.” Pratt explained that it’s fairly simple to predict the erratic behavior of, for instance, a cyclist, and program an autonomous vehicle with a suitable response. The tough part is creating a response to the things we can’t anticipate.
“Imagine debris falling off a truck,” says Pratt. “Should the car think of the debris like another car, and avoid it? Kind of, but the debris might break apart, so then it’s like lots of cars. Should it think of it as a pedestrian? Kind of, but the debris might initially be moving far faster than a pedestrian. These are really hard problems. Part of our work will focus on augmenting machine learning, and measuring the robustness of systems for scenarios we haven’t yet thought of.”
The second TRI autonomous project is called “The Car Can Explain.” As Pratt puts it, “autonomous cars have the power to make life or death decisions, so they must be capable of explaining those decisions. When the car does something unexpected, it needs to tell us exactly why that happened. Cars will become more intelligent in that way.”
According to Pratt, Toyota now has millions of miles of autonomous reliability under its belt, but he says the company needs to get to trillion-mile reliability, based on Toyota producing 10 million vehicles a year, each covering around 10,000 miles annually and staying on the road for a decade: So in a typical year, Toyotas travel approximately a trillion miles worldwide.
“We may be 95 percent of the way to full autonomy, but that doesn’t mean the last 5 percent will be as easy as the first 95 percent. It’s like climbing a mountain, where the final ascent is the hardest. And if just a small percentage of the trillion miles that Toyotas cover annually is very difficult driving, it’s still many, many miles; that’s the part we need to address.”
Luckily, Pratt has brainpower on his side, Toyota describing the team as “an all-star group of scientists and engineers.”
Photo: Pratt at CES
Top TRI hires include:
• Larry Jackel, machine learning expert and a former Bell Labs department head and DARPA program manager
• Eric Krotkov, COO and former DARPA program manager
• James Kuffner, area lead in cloud computing, ex-head of Google Robotics
• John Leonard, MIT professor and autonomous driving expert
• Hiroshi Okajima, executive liaison officer and Toyota R&D manager
The advisory board includes Daniela Rus, MIT’s computer science and AI laboratory director and Bran Ferren, former president of Walt Disney Imagineering’s R & D.
TRI is also embracing a collaborative environment, inviting suppliers, rivals, academics and others to contribute, while sharing its ideas widely.
“We don’t have not-invented-here syndrome and we shared our 5,600 hydrogen patents,” says Pratt. “At Toyota, we believe when good ideas are shared, great things can happen.”