Building Smart Robots
“A personal robot could save billions of dollars in elder care while letting people stay in their own homes and communities."
- Rod Grupen
Speech language pathologist Yu-kyong Choe (Communication Disorders) is working on a two-year grant funded by the American Heart Association to investigate the effect of stroke rehabilitation delivered by a humanoid robot, a child-sized unit with arms and a screen where therapists, doctors and others can interact with a client. Choe is collaborating with Rod Grupen, director of the Laboratory for Perceptual Robotics, on ways to bring more and longer-term, home-based therapy and social contact to people recovering from stroke.
It’s estimated that 3 million Americans daily experience the debilitating effects of stroke. But even after years, they can recover significant function with intensive rehabilitation, says Choe. The bad news is that this is rarely available or accessible due to a shortage of therapists and lack of coverage for long-term treatment. Many people are left with chronic low function, which can lead to social isolation and depression.
The researchers acknowledge that some may object to robots delivering therapy, but they say the need is great and definitely not being met now, especially in rural areas. The goal is to aid human-to-human interaction, so a robot can temporarily take the therapist’s place and even help with routine tasks. Grupen says, “In addition to improving quality of life, if we can support a client in the home so they can delay institutionalization, we can improve outcomes and make a huge impact on the cost of elder care. There are 70 million baby boomers beginning to retire now.”
Choe adds, “In this study, we want to find out whether the physical presence of a robot that looks like a human can make a person more motivated to complete her therapy tasks. Also, whether she’ll engage and show progress with the robot, because intensity really matters. The more you practice, the better function will be. Anything that keeps her from being bored or quitting is valuable.”
“Thirty years ago when computers were massive and bulky, no one thought they would be in our homes,” she points out. “If the same is true for robots, now is the time to start developing home robots that effectively lead language and physical therapy routines for people recovering from stroke. We want to support this with sound research.”
The study will enroll five stroke patients per year to attend three sessions per week for five weeks at the UMass Amherst lab. Three treatments will be compared: computer-mediated, robot-mediated and robot-assisted telepractice by a remote therapist.
In the robot-mediated condition, patients complete word-retrieval tasks and games, plus arm exercises, delivered by the robot alone based on therapy routines it has observed. In the computer-mediated condition, the same tasks and exercises will be presented on a laptop computer.
In the robot-assisted telepractice condition, the client performs word- and arm-movement tasks designed and directed by a therapist in a remote location being observed and mimicked by the robot via 3-D range camera. The robot exactly copies the therapist’s movements. Choe predicts that the two robot-mediated conditions will yield better outcomes in both speech and physical function because of the patient-robot interactions. The research team will also analyze how the telepractice and robot-mediated therapy sessions are received by client and therapist.
Therapists Jennifer Baird and Tammie Foster, with computer science doctoral students Takeshi Takahashi and Hee-tae Jung, are working with patients three days per week and developing software for uBot5, an adaptive humanoid robot, to act as the liaison between a remote therapist and the client at home.
Jung, who evaluates how well the robot is learning therapy routines and goals, says early robots tended to be too task-specific, but now scientists are designing more flexible, adaptable robots. “I want the robot to pick up some specific skills and to facilitate interaction with flexible behavior. I also want to know whether the therapist feels it’s a valuable tool and whether patients like it,” he says.
Grupen explains, “We hope to advance artificial intelligence by creating robots that learn from human beings while interacting with them. It will also allow doctors, therapists and family to come into the home via telepractice. We want to explore embedding robots into human culture in a way that improves quality of life and increases human-to-human social interaction in a circumstance when age and disability can isolate people.”
“Stroke rehabilitation is such a monumental financial problem everywhere in the world, that’s where it can pay for itself,” he adds. “A personal robot could save billions of dollars in elder care while letting people stay in their own homes and communities. We’re hoping for a win-win where our elders live better, more independent and productive lives and our overtaxed healthcare resources are used more effectively.”