A New Intelligent Approach to Patient-cooperative Control of Rehabilitation Robots

This paper presents a new intelligent method to control rehabilitation robots by mainly considering reactions of patient instead of doing a repetitive preprogrammed movement. It generates a general reference trajectory based on different reactions of patient during therapy. Three main reactions has been identified and included in reference trajectory: small variations, force shocks in a single moment and variable level of participation. This reference trajectory can facilitate every kind of patient-cooperative controller of rehabilitation with an evaluation factor. A simple planar 2-DOF exoskeleton is used as rehabilitation robot for leg therapy. The low level controller used here is Brain Emotional Learning Based Intelligent Controller (BELBIC) which is participating to develop an admittance control scheme. Performance of this controller has compared to PID with simulations and better results have observed specially in control signal and speed of reaction when controller encounters a force shock.