First implementation results on FAT based adaptive control for a lower extremity rehabilitation device

Abstract

Gait restoration after stroke is a major concern of neurological rehabilitation. In this paper, a recently developed wearable lower extremity rehabilitation device is introduced, and an adaptive control strategy based on Function Approximation Techniques (FAT) is designed to aid stroke patients who are in the early phase of rehabilitation or severely affected. The device adopts an anthropomorphic structure actuated on hip and knee joint level in sagittal plane. Frameless DC motor with high gear-ratio harmonic drive is used as actuator. Fourier series fitting method is used to obtain the desired hip and knee reference trajectory. This control algorithm is then implemented in the device with friction compensation to track the desired gait trajectory. Implementation results for a healthy subject show that good tracking performance can be achieved, and actuator toques are within a reasonable range. This FAT based adaptive control algorithm could potentially be used for stroke patients as well.