IEEE Life Fellow Takeo Kanade is no stranger to high-profile engineering projects. He has helped develop robots for military, space, and entertainment applications. But during the past decade, he turned his attention to what he calls the less glamorous but more fulfilling development of quality-of-life robots.
Kanade, a professor at the Carnegie Mellon University Robotics Institute, in Pittsburgh, is the founder of the university’s National Science Foundation Engineering Research Center and served as its director from 2006 until last year. The ERC develops quality-of-life technologies such as robots that perform household chores. Kanade was guest editor for the August 2012 Proceedings of the IEEE special issue on the subject.
His specialty is computerized vision recognition—software and hardware that enable computers to “see” by electronically acquiring, processing, and analyzing images. Kanade’s 40-year career has focused on several areas of robotics, including manipulators, sensors, and autonomous mobile robots, as well as multimedia applications that rely on a large number of cameras to model the environment.
His work overseeing ERC projects has made him revise his whole approach to robotics, he says: “In the past, robots were designed for autonomy in manufacturing, space, and military industries. The idea was to reduce human involvement and, by doing so, decrease production costs and increase safety. Quality-of-life robotics is the opposite. Humans are now part of the system. But humans are the most difficult and least understood part of it—that’s the challenge.”
It’s an interdisciplinary research area. Clinicians and caregivers are involved, describing patient needs and giving feedback on prototypes. “Instead of starting out with engineers saying, ‘Let’s build this cool robot and see if people like it,’ you need to start with what people want and how they’ll accept working with a robot,” Kanade says.
A HELPING HAND
Kanade has overseen a variety of projects at the ERC, including cellphone apps that tell you when the next bus is coming and robots that prepare meals. The Home Exploring Robot Butler is a two-armed robot on wheels that can pick up objects, like cups and milk cartons. The Personal Mobility and Manipulations Appliance is a wheelchair with arms that can be controlled by the rider or remotely by a caregiver.
First-person vision is another ERC research area. One project applies a wearable camera that captures the point of view of the user to better understand the person’s interaction with the environment and determine whether cognition and motor skills are deteriorating. Camera data is sent to a computer, which analyzes the images. Once the system understands what the person is doing and needs, it can offer the appropriate advice by voice or display.
Kanade discovered his calling as an undergraduate electrical engineering student at the University of Kyoto. He found a mentor in Toshiyuki Sakai, a professor at the university and a pioneer of real-time speech recognition. “I thought, ‘If we can teach a computer to recognize speech, why not make it visually recognize objects such as faces?’” Kanade says.
Graduating in 1968, he continued at Kyoto for his master’s degree (in 1970) and Ph.D. (1973) in electrical engineering, pioneering a computer face-recognition system. He then joined the university as an assistant professor of information science during an exciting time in his country’s technological history. “It was the dawn of computer science in Japan,” he says. “Computers were moving from number-crunching machines to ones used for what was then called non-numerical information processing, or what today is called multimedia.”
After a guest-researcher stint at Carnegie Mellon, in Pittsburgh, in 1977, Kanade joined its Robotics Institute in 1980, where he focused on autonomous systems. In 1984, he initiated and led the Defense Advanced Research Projects Agency–funded Autonomous Land Vehicle project to develop a self-driving, all-terrain vehicle with computers, cameras, and 3-D sensors on board. Kanade’s group also developed 3-D vision systems for planetary exploration, as well as a robot that moves along trusses like an inchworm.
He gained some notoriety when his project, EyeVision, an instant-replay system inspired by the movie The Matrix, was developed for CBS TV and used in its 2001 Super Bowl coverage—earning him an interview during the network’s game coverage. “I’m the only professor who ever appeared on the Super Bowl,” he says, laughing. He later had a cameo in Surrogates, a 2009 science fiction movie starring Bruce Willis. The movie used archived TV footage of Kanade talking about robots becoming smarter than humans.
By 2000 he began thinking about robotics in more symbiotic terms and, five years later, he landed a US $5 million annual grant from the NSF to start the ERC. The grant ends in 2015, by which time the center is expected to be self-sustaining through patent royalties and other funding.
An impetus for Kanade’s focus on quality-of-life technologies was his aging mother’s rapid decline after an accident that significantly decreased her mobility. Kanade was visiting her frequently in Japan, and he wanted to monitor her between visits. He wondered whether robotics could help. “My mother died before I got the grant,” he says, adding that her spirit propels his work.
“As a young engineer, I think I was romantic about the future of robotics,” he says. “I believed that robots and computers would be smarter than humans. Now, I believe that robots and humans enhance each other’s performance. I see them as having a beautiful friendship.”