Robots That Care

By Jerome Groopman

"Born in Belgrade, in what was then Yugoslavia, Maja Matarić originally wanted to study languages and art. After she and her mother moved to the United States, in 1981, her uncle, who had immigrated some years earlier, pressed her to concentrate on computers. As a graduate student at the Massachusetts Institute of Technology, Matarić wrote software that helped robots to independently navigate around obstacles placed randomly in a room. For her doctoral dissertation, she developed a robotic shepherd capable of corralling a herd of twenty robots. At the end of her graduate training, Matarić, influenced by her knowledge of cognitive science, became interested in how people could benefit from interacting with robots. Now forty-four and a professor of computer science at the University of Southern California, she has begun working with stroke and Alzheimer’s patients and autistic children, searching for a way to make machines that can engage directly with them, encouraging both physical and cognitive rehabilitation.

“We wanted to do something entirely different,” Matarić told me. She assembled a team of experts in several disciplines: psychology, mechanical engineering, kinesiology, rehabilitation medicine, and neurology. The team members observed Isaac Asimov’s First Law of Robotics: the robot must not injure the patient. They also had to determine what tone of voice was optimal, what type of language the robot should use, how close it should get to the patient—essentially, what kinds of personality and temperament were most effective, and for what kind of patient. The robot would coach the patients orally, rather than physically. (One that physically touched a patient might require approval by the Food and Drug Administration as a device, given the potential safety issues.)