An Assist-as-Needed Controller with Fault-tolerant Region for Safe and Effective Training on End-effector Upper Limb Rehabilitation Robot

This paper proposes an assist-as-needed position-based impedance controller for rehabilitation training of upper limb motor function impairments in patients. The controller establishes a fault-tolerant region around the movement training trajectory, allowing the upper limb to move freely in this space without assistance from the robot. For the movement out of the fault-tolerant region, the robot provides compliant assistance to guide the upper limb back toward the movement training trajectory. This approach minimizes assistance from rehabilitation robots and maximizes the patient's active participation, which could promote rehabilitation efficacy. This study not only simulates the robot arm but also established an equivalent model of the upper limb coupled with a robot model and implemented a human-robot interaction system for control simulation that enhanced the reliability and practicality of the rehabilitation training. The simulation results possess enhanced reliability and practicality, and to some extent, could validate the performance for safe and effective rehabilitation training.