Gravity compensation improvement for a cable-actuated elbow rehabilitation robot using sliding mode control

Performance increasing of robot-aided training in stroke elbow rehabilitation is the goal of this paper. Therapist holds on the arm of patient and guides the center of mass along a desired trajectory. In robotic rehabilitation, when the arm of patient rotate within the desired boundaries, (s)he should ideally not feel the robot. The robot needs to actively compensate for the weight of the exoskeleton and reflected mass of the motors. A nonlinear torsion spring can be used and also a counter-torque as a function of arm angle is applied by the motor. Applying the springs affords more convenience, it allows smaller motors to be used, the size of required brakes can be reduced and inherent safety is introduced in rehabilitation robots. Furthermore, the robust controller design can be used to compensator the modeling errors and gravitational force. A novel elbow rehabilitation robot is designed based on the cable actuation. The strategy is not just anti-gravitational forces because there should be joint-stiffness control. The uncertainty in the patients arm dynamic is effectively approximated. The motion of closed-loop control system in the presence of parametric uncertainties is investigated. The sliding mode controller with proportional-derivative controller is compared through computer simulation and improvement is observed.