Abstract: Aiming at the difficult problem of nonlinear solution in repetitive rehabilitation training of multi-joint upper limb exoskeleton, a closed-loop PD iterative learning control method is proposed. The parameters, degrees of freedom configuration and joint motion range of the six-DOF upper limb exoskeleton rehabilitation robot are determined based on ergonomics. The proposed exoskeleton uses the human-robot interaction force as the coupling method, a human-robot coupling model based on the Newton-Euler method was established, and the human-robot coupling dynamics simulation analysis is complete. Based on the iterative learning control theory, a closed-loop PD iterative learning control method for exoskeleton rehabilitation robot is proposed. The trajectory error, human-robot interaction force and driving torque of the iterative learning control of the shoulder/elbow joint were analyzed through modeling and simulation. The trajectory error after the third iteration is less than 0.05 rad, and the output of the PD iterative learning controller effectively compensates the system control and improves the stability of the system state. A prototype of six-DOF upper limb exoskeleton rehabilitation robot was developed and the relevant tests were performed. The experimental results show that as the control experiment runs in the iterative domain, the output of the system transforms to the desired system state. The proposed iterative learning control algorithm can improve the control accuracy of the repetitive motion of upper limb exoskeleton rehabilitation training, thereby improving the performance of human-computer interaction.

Key words: upper limb exoskeleton, human-machine coupling, rehabilitation training, iterative control met hod