Trajectory Optimization by Tacit Learning

To support the rehabilitation of the paralytic by using the robots, the controller must be adapted to the degree of recovery of the patient. Tacit learning that can create the robot behavior adapted to the environment is applicable to create the support robot behavior depending on the state of paralysis. In this paper, the trajectories optimization depending on the weight attached on the end-effector is discussed to find the potential of tacit learning to apply it for rehabilitation support. The simulation and experimental results showed that the trajectories of the 3DOF manipulator were optimized without using any supervising signals and cost functions, but using body/environment interactions. These results suggest that tacit learning can control the support robots depending on the paralysis states of the patients.