Toyota Staffs Up for AI, Robotics Research

The rising importance of robotics on the global stage was thrown into stark relief last fall when Toyota Motor Corp. set up a new U.S.-based company and said it would spend $1 billion over five years on research and development of autonomous cars and other robotics products.

Now the new Toyota Research Institute (TRI) is forging its way into a challenging recruiting environment as it seeks to hire some 200 experts for two new campuses, one on Stanford University land in Palo Alto, Calif., and another next to the Massachusetts Institute of Technology in Cambridge, Mass.

Toyota’s bold maneuver came as robotics startups are proliferating and other companies large and small are bulking up their own benches with experts in a number of related fields, such as artificial intelligence, machine learning, computer vision, cloud computing, computer simulation and control, and predictive analytics.

Major automakers, including Ford, General Motors, BMW, Hyundai, Mercedes, Nissan and VW/Audi, have also been beefing up their research and development efforts in Silicon Valley.

‘Demand outstrips supply’

“Every robotics company I know is hiring, and there is stiff competition from large companies that you don’t traditionally think of as robotics companies,” said Andra Keay, director of Silicon Valley Robotics, an industry association.

Some companies, such as Google Inc., are hiring robotics experts and putting them on non-robotics computer science tasks. “Demand certainly outstrips supply,” she said.

Straight robotics job listings rose by 1.4 percent per 10,000 posts between January 2014 and January 2016, from 4.6 percent to 6 percent, but jobs in the related areas of AI and machine learning rose by about 5 percent, from 7.3 percent to 12.2 percent, according to the jobs website Indeed.com.

The money and prestige flowing from the world’s largest automotive company and its stated mission are clearly influential, however, as evidenced in September by the hiring of Gil Pratt, former project manager of the DARPA Robotics Challenge, to be TRI’s CEO.

In addition, TRI in January introduced Pratt’s nascent leadership team, which includes former Google Robotics head James Kuffner, who is credited with coining the term “cloud robotics.” Team members were in Japan at the end of January for a series of meetings with Toyota executives to set priorities.

“They are running like mad,” said John Hanson, who leads Toyota’s U.S. advanced technology and business communications. “It is still in an organizational phase.”

Toyota is looking to plumb the expertise at both MIT and Stanford, both for full-time positions and part-time consultants, and it is financially supporting research at both institutions.

“The huge hiring demand coming from large companies that can afford to pay really well is a big problem for the smaller robotics companies,” Keay said. “These giants wield a big checkbook, and they have great recruitment agencies.”

One Stanford alumnus who recently joined TRI full-time in Palo Alto is Michael Sherman, a former bioengineering researcher who is now TRI’s program manager of simulation software for development of safe controllers. According to his LinkedIn profile, Sherman will be working on Toyota’s autonomous vehicles and home-assistance (“partner“) robots.

Sherman previously was chief software architect for Stanford’s Simbios Center, which is funded by the National Institutes of Health and focuses on simulating biological structures.

Toyota investments support AI vision

The $1 billion Toyota said it would spend on TRI over five years was in addition to $50 million it committed last fall to the effort, which is being led by Daniela Rus, director of MIT’s Computer Science and Artificial Intelligence Laboratory, and Fei-Fei Li, director of the Stanford Artificial Intelligence Laboratory.

The MIT team will focus on making autonomous systems accountable, while Stanford’s will focus on developing machine systems for making decisions in unanticipated situations.

“Our research is aimed at improving mobility and transportation by advancing the science of autonomy and pushing the envelope of robotics, computer vision, and artificial intelligence,” Rus told Robotics Business Review.

Rus said she is excited to reduce traffic fatalities with autonomous vehicles, as well as to increase mobility of people in their homes — something that Toyota has said presents enormous opportunities, given the large number of people entering old age.

Toyota is working on its Human Support Robot.

Toyota last year said it was ramping up development of its Human Support Robot, a one-armed machine on wheels that can help older and sick people by picking up and carrying objects.

Maryanna Saenko, an autonomous systems research analyst at Lux Research Inc. in Boston, said that while other automotive companies, including Honda, have done notable research in robotics, Toyota’s statements about TRI struck a novel tone.

“Toyota is saying, ‘We might not make cars one day,'” Saenko said. “Toyota is being a little bit bold. It has been a while since we have heard a bold vision statement from an auto company.”

“Gil [Pratt] is pulling the world’s top academics and their top students into what might be the next Bell Labs,” Saenko said.

“They went and hired the people who actually may make the vision a reality,” she said. “This pooling of talent I think is vastly exciting.”

Looking beyond cars to robots

Hanson similarly emphasized that Toyota is treating its investments in AI and robotics as though they could be the future of the company, likening the present moment to the 1930s, when Toyota shifted its business model from making fabric looms to manufacturing automobiles.

Automobiles are obviously not going anywhere in the near future, but “home robotics is going to be a major focus” for TRI, Hanson said.

“We may be entering a new phase of the company,” Hanson said.

TRI’s initial technical team includes:

Former Bell Labs department head and DARPA program manager Larry Jackel, who will work on machine learning
Former DARPA program manager Eric Krotkov, chief operating officer
MIT Professor John Leonard, who will work part time in the field of autonomous driving
MIT Associate Professor Russ Tedrake, who will work part time in the field of simulation and control

TRI’s advisory board includes big names from a variety of backgrounds, including:

Emeritus MIT Professor Rodney Brooks, former director of the MIT Computer Science and AI Lab; founder of iRobot Corp.; and founder, chairman, and chief technology officer of Rethink Robotics Inc.
Richard Danzig, former secretary of the U.S. Navy and an expert in cyber security
Yann LeCun, director of AI at Facebook Inc.
John Roos, former CEO of Wilson, Sonsini, Goodrich & Rosati and former U.S. ambassador to Japan. He is currently general partner at Geodesic Capital, a late-stage venture capital firm, and senior advisor at Centerview Partners, a mergers and acquisitions advisory firm. Roos is an expert in technology innovation in Silicon Valley.

About the author:
Patrick Hoge is a veteran news reporter who has written extensively on government and business, including most recently about the San Francisco Bay Area’s technology scene for the San Francisco Business Times. His writing has appeared in the San Francisco Chronicle, The Sacramento Bee, and Robotics Business Review, among other publications.