Iterative Impedance Learning Control for Ankle Rehabilitation

In this paper, impedance learning control is investigated for conducting robot-aided ankle rehabilitation. Under repetitive interaction tasks, the ankle dynamic is described as a time-varying iterative system with unknown mechanical impedance parameters. The gradient following approach and iterative learning algorithm are employed to obtain a desired impedance model. With learned parameters, an inner torque controller with robot dynamic compensation is implemented for tracking the modified trajectory. Experimental results with an ankle rehabilitation robot prototype validate the efficacy of proposed method.