Semi-autonomous Robot-assisted Cooperative Therapy Exercises For A Therapists Interaction With A Patient
Recent increases in demand for rehabilitation therapy combined with limited therapist time and patients living in remote areas have created a significant burden on healthcare delivery worldwide. Techniques allowing time sharing a therapist between multiple patients using semiautonomous techniques such as Learning from Demonstration (LfD) based robotics are an attractive solution to address this growing demand for rehabilitation. A Gaussian Mixture Model (GMM) and Gaussian Mixture Regression (GMR) based approach to LfD is demonstrated with the purpose of creating a versatile system to assist with therapy exercises in the absence of the therapist. A bilateral telerehabilitation system is used to enable collaborative performance of a one Degree of Freedom (DOF) task by a therapist and a patient. We demonstrate that the learned model of behaviors is versatile enough to respond appropriately to a variety of patient actions.