An exciting new interdisciplinary research project at UNH could lead to intelligent robots that guide patients through therapeutic exercise training. The project was recently awarded $750,000 by the National Science Foundation’s Division of Information and Intelligent Systems and is led by Momotaz Begum, assistant professor in the Computer Science department, along with Dain LaRoche, associate professor of Exercise Science in the department of Kinesiology and Sajay Arthanat, associate professor of Occupational Therapy.  “This NSF grant will help us to design robust robot learning algorithms that will be tested on real users for exercise training. Such interdisciplinary efforts are a key to bring learning-based robotic systems out of the laboratory and employ them in the service of humans,” states Professor Begum.

The research project is designed to develop intelligent algorithms that allow robots to learn new exercises directly from demonstrations by a therapist and helps address a significant shortage of physical and occupational therapists anticipated over the next 20 years. The intelligent robots are not programmed, but rather utilize a new “Learning from demonstration (LfD)” framework. LfD enables a robot to function in a teaching capacity for patients, as well as quantitatively evaluate a patient’s performance of the training exercises. “For robot therapists to be effective, they must be able to model an exercise accurately, assess patient performance of the exercise and provide correction in real time, and be accepted by patients and clinicians,” said Professor LaRoche.

This collaborative project brings together strengths across three different disciplines and two colleges (the Colleges of Health and Human Services and Engineering and Physical Sciences) at UNH and reflects an institutional commitment to supporting interdisciplinary collaboration. Initial funding for the research came from the Collaborative Research Excellence (CoRE) initiative, an internal grant program now in its second year.  According to LaRoche, “The CoRE program allowed us to assemble a research team with the complementary skills needed to develop and evaluate therapeutic robots, it mitigated the barriers of collaboration, and supported an initial research study that was the foundation for the NSF grant application.”