Robots in Rehab

Specialized robots being built by a team of members are able to help deliver much therapy

5 December 2008
tech The Manus robot helps a stroke victim use her disabled hand to move targets on the screen.

The good news: contrary to what many clinicians say, the human brain exhibits considerable plasticity. It can change its internal structure, particularly its synaptic connections, throughout its lifetime, not just during the formative years. That means it is possible to ameliorate the effects of a stroke, cerebral palsy, and other afflictions caused by brain damage.

The bad news is that those beneficial changes require prolonged sessions of highly repetitive movements that are difficult for therapists to perform. Adding to the problem is that insurance providers apparently believe, incorrectly, that recovery is possible for no more than a few weeks after an injury such as a stroke, and they are willing to pay for only short periods of therapy.

However, specialized robots being built by a team at MIT are able to help deliver much of the therapy. What's more, the robots have proven their effectiveness in reducing elbow and shoulder impairments in stroke victims, according to a recent article by IEEE Senior Member Hermano Igo Krebs, researchers Bruce Volpe and Neville Hogan, and others. The article, "A Paradigm Shift for Rehabilitation Robotics," appeared in the July/August 2008 issue of IEEE Engineering in Medicine and Biology Magazine. In the opinion of Volpe, a professor of neurology and neuroscience at Weill Cornell Medical College, in New York City, when robots have demonstrated an inarguable ability to "bring your wrist and hand back, then it will be something that everyone will clamor for, and the [attitude of insurance providers] will change."

RETHINKING REHAB In stroke patients, neurons in the brain die from lack of oxygen. The resulting neurological deficit has long been thought irreversible. But researchers such as Volpe, working with MIT mechanical engineers, have demonstrated that reversing that neurological deficit in stroke patients is possible even after many years.

"The brain is a learning machine. Just because it is damaged doesn't mean it can't learn," Volpe says. "It learns with the kind of entrainment that a robot can deliver: prolonged, highly reproducible, high-intensity, interactive therapy."

The robot he is talking about is the MIT-Manus, developed by Krebs, a principal scientist in MIT's mechanical engineering department. Working with Neville Hogan, professor of mechanical engineering and brain and cognitive science at MIT, Krebs designed the robot specifically to work with patients interactively.

With Manus, a patient's dysfunctional arm is strapped to the robot's arm, the fingers wrapped around a cylinder, and the patient is asked to play a video game in which she moves a cursor from point to point on a computer screen by trying to move her arm and hand [see photo]. If she does that successfully, the robot does nothing but monitor her actions. But if she cannot, or is too slow or wanders too far off-course, the robot assists her by moving the hand and arm for her or by resisting off-course movements, giving the patient the sensation of coming up against a spongy wall. The farther she deviates from the desired path, the harder she has to push. As Hogan puts it, the spongy wall doesn't stop you from deviating from the nominal movement, but it discourages it. A critical feature of the system is that the amount of assistance and the degree of challenge (for example, how the robot defines "too slow") varies with how well the patient is doing. If the patient does well, the assistance decreases while the challenge increases.

Movements are repeated many times during a therapy session—up to 1200 movements in an hour. A human therapist offering similar hand-over-hand assistance is lucky to be able to do 50 or 60 movements per session, according to Volpe. He says that advantage is a major reason for the success of the robotic approach.

ENGINEERING RECOVERY The big challenge in building Manus is to have it move a patient's arm to a desired position while remaining unobtrusive when it's not needed. That is, its mechanical output impedance should be as low as possible. According to Hogan, the impedance depends mainly on the rotational inertia of the robot's motor.

For their next generation of robots, Hogan and his team at MIT are working on a controller that senses when the patient is applying a force to the robot arm and then tells the motor to drive the arm in the direction of the force. Such an active control scheme could lead to robots capable of generating forces greater than body weight while still maintaining a feather-light touch for the patient.

Krebs is studying how to adjust treatment variables to produce the best outcome. He found that treatments of one hour per day, three days a week for six weeks, produce much better results than a more intensive regimen of two hours per day, three days a week for three weeks, even though the total number of hours is the same. He says such issues will be the main focus of his research for the next 10 years.

FOR MORE INFORMATION about this topic, download the abstract of the cited article at

IEEE membership offers a wide range of benefits and opportunities for those who share a common interest in technology. If you are not already a member, consider joining IEEE and becoming part of a worldwide network of more than 400,000 students and professionals.

Learn More