Performance-Based Iterative Velocity Updating Method for Robot-Assisted Upper Limb Training in Reaching Motions

Abstract

This paper presents an Iterative Velocity Profile Updating Controller (IVUC) to improve robot-assisted upper limb rehabilitation by dynamically adjusting velocity profiles during robotic exoskeleton training sessions. The proposed approach personalizes motion assistance by iteratively updating bell-shaped velocity profiles based on the subject's performance. Key parameters, including the maximum value axis (adjusted via a time shift function), total displacement, and time duration, are tuned using a discrete proportional derivative iterative learning controller (PD-ILC). The updated parameters generate optimized trajectories for subsequent training iterations. Two experiments validate the method: the first identifies the optimal time shift function, while the second evaluates the approach with subjects performing a predefined reaching motion using the TAMU CLEVERArm exoskeleton. The results show that iterative trajectory updating reduces both the number of iterations required to reach steady-state performance and the final tracking error, highlighting its potential to improve rehabilitation efficiency.