Learning And Robotic Imitation Of Therapist’s Motion And Force For Post-disability Rehabilitation

The inclusion of robots in therapy is becoming common due to robots’ power, repetitive motion ability, reprogramming capacity and adaptability to new tasks. In recent years, the demand for rehabilitation services has increased due to the rising number of patients with disability. In this paper, we propose a solution to the rising demand for therapists’ services by combining Learning from Demonstration (LfD) and robotic rehabilitation. The goal of the paper is to implement LfD to model and learn the therapist’s behavior (be it a trajectory or force) as a nonlinear dynamic system using a method called Stable Estimator of Dynamical Systems (SEDS) to later reproduce the learned behavior in the absence of the therapist using a robot. This method allows the therapists to first train a robot to learn his/her behavior such that, later when the therapist is no longer involved and the patient works alone with the robot, the robotic system determines whether and how to interact with the patient the same way the therapist would have interacted.