Toyota's A.I. research efforts could mean cars that anticipate traffic, pedestrian moves
Last week, the auto maker announced that it was investing $1 billion over the next five years to establish two research and development centers – the main center in Silicon Valley and a secondary campus near MIT in Massachusetts. The project is expected to eventually include 200 scientists and engineers.
The Toyota Research Institute, which is slated to open in January, will be led by Gill Pratt, who until recently was a high-ranking leader at DARPA, heading their Robotics Challenge, a competition aimed at advancing humanoid robots. The research will initially focus on how people and machines can work together, specifically in the area of mobility.
Last week's announcement was just the latest A.I. research news to come out of Toyota.
In September, Toyota announced a separate partnership with Stanford University and MIT to push research into A.I. and robotics.
In that effort, Toyota is contributing $50 million over five years – half to Stanford and half to MIT -- to establish research centers that will focus on A.I. and in advancing human and robotic interactions.
By combining research on the interaction between humans and machines, artificial intelligence and big data, Steve Eglash, executive director of the artificial intelligence lab at Stanford, is hopeful that Toyota's richly funded research centers will, in the next five years, lead to cars that operate more safely in bad weather and on city streets.
He also hopes their work will lead to robots that can help the elderly and sick live safely, and happily, in their own homes.
"This isn't just about creating scholarly results and publishing them in scholarly journals, it's about having an impact and changing the world," said Eglash, who is heading the Toyota-funded research center at Stanford. "This is the same culture that brought things like Google and Yahoo and HP and so many others. This is a group that is really good at focusing on what needs to be done."
Stanford's research center, which will be housed in a pre-existing building on the campus, will work on seven to nine projects under the Toyota-funded effort, Eglash said.
Each research project will have two to three university professors, along with two to five graduate students focused on it.
Eglash said preparations began a few months ago to identify what research projects they'll take on, to put out a call for proposals and to launch a review and selection process.
"Toyota provides additional resources and provides exciting focus in some of the interests," he said. "It begins with the financial resources but Toyota brings a whole lot more to the party than just financial support. They have a unique perspective on the future of the A.I. industry and robotics."
Increasingly, A.I. research is about more than algorithms and smart systems. It's also about big data, and Toyota has plenty of data on automobiles and how people interact with them.
"Data is an important resource for a whole lot of research and Toyota brings us a whole lot of data," explained Eglash. "We were all used to thinking of an organization's resources as their plant and capital resources, but increasingly we also think of data being a valuable resource. Cars are one of the most connected things we own. How can we use that intelligence How do we understand how people want to interact with their cars That's an important part of this."
Eglash also hopes their efforts will make artificial intelligence more contextual and human-centered.
"When your computer is providing you with information, whether it's a computer, a phone or a robot, you want it to be as aware of the situation you're in as possible," Eglash said. "You want the car to understand if your mind is elsewhere and you want a more passive experience. Maybe you're old and your eyesight isn't as good and you need more help. In the future, we'll want intelligent machines to interact with us the way we want them to interact with us."
So how long before cars recognize if a driver is tired and might need help staying focused on the road, or might even take over the driving
It won't be long at all, according to the Stanford researcher.
The auto industry already has introduced cars that can park themselves, stay within the lane and brake when they sense an object in their path.
Hopefully in a few years, cars will make predictions about traffic and road conditions in an area minutes before the car even gets there. It's also hoped that cars will soon be smart enough to anticipate what bicyclists and pedestrians might do and take precautionary maneuvers.
"We have the potential to use technology to drastically reduce the number of accidents," Eglash said.
While the majority of Stanford's research will focus on using A.I. in automobiles, researchers also will work on human/robot interactions.
For that, Stanford will pull in professors from the humanities, psychology and social sciences to work with the program's engineers.
"It's not just about technology," Eglash said. "We don't know what will happen, but we believe that research works best when you have as diverse a group of disciplines as possible."
This major commitment to A.I. research is exciting for scientists trying to move the science forward, according to Manuela Veloso, a computer science professor at Carnegie Mellon University, which is not part of the Toyota project.
"It's exciting to see that people are realizing how much the field of A.I. can make a difference in people's lives," she said. "The work they'll be doing on cars can be used elsewhere. We need better decision making based on an understanding of a situation. Machines need to understand if someone is crossing the road or crossing a room. This is the beginning. It's the reality of A.I. in the physical world, whether it's cars or robots that move indoors or in supermarkets."