Abstract: In order to synchronously realize the overshoot suppression of joint trajectory and variable strength training in the human-computer interaction compliant control of exoskeleton rehabilitation robot, an internal and external loop nested adaptive impedance control method with controllable damping is proposed, in which a shoulder-elbow two-degree of freedom upper limb rehabilitation robot is adopted as the research object. Firstly, a joint trajectory tracker as an inner loop controller is designed, in which the electromagnetic damping torque is introduced into the dynamic equation of the rehabilitation robot through the structural mode output by the magnetorheological damper and the joint motor in parallel. It takes angular velocity and angular acceleration vectors of moving joints as inputs and calculates the excitation current value and the joint torque gain coefficient to adjust the damping moment to suppress the overshoot. Secondly, an adaptive trajectory generator as an outer loop controller is designed by utilizing Lyapunov functions, in which the human-computer interaction force, joint angular velocity and angular acceleration are employed as input to real-time computing and correction of the desired trajectory, and it by adjusting the fighting stiffness of the model parameters to realize the human-computer interaction force and adaptive strength training. Finally, numerical simulation experiments show that compared with the control method without controllable damping, the overshoot of shoulder joint and elbow joint under the proposed method are better than 96.2% and 97.5% respectively, and the variable strength training is realized only by adjusting the stiffness parameters, which verifies the feasibility and effectiveness of the proposed method.

Key words: controllable damping moment, self-adaptive, active rehabilitation training, impedance control