Performance-based Adaptive Control Strategy for Robot-assisted Rehabilitation

A performance-based adaptive control strategy is proposed for robot-assisted upper-limb coordinated rehabilitation training, with aims to provide subject-specific robotic assistance for patients with limb movement disorders and promote their active engagement in effective rehabilitation training. Firstly, an end-effector for upper-limb bilateral rehabilitation and a coordinated training task are presented. Then, kinematic parameters based on clinical scales and coordinated training indicators are introduced, and a fuzzy neural network model is constructed to relate the multiple task measures to the robot admittance parameter. Finally, the coordinated training experiments are conducted with human subjects to validate the proposed method and compare it with the human-robot interactive strategy mentioned in the previous study. Experimental results indicate that the proposed strategy achieves a good ability of tracking desired performance indicators and keeping human-robot interaction smooth, and it can adaptively provide tailored robotic assistance, which enhances subjects' active engagement in training.