A Novel Assist-As-Needed Controller Based on Fuzzy-Logic Inference and Human Impedance Identification for Upper-Limb Rehabilitation

Abstract

Upper-limb rehabilitation robots are being increasingly included in the rehabilitation after stroke. Clinical studies suggest that the patient's active participation can maximize effectiveness of robot-assisted rehabilitation and result in fastest possible recovery. In order to promote active involvement of the patient, we develop a subject-adaptive controller within the assist-as-needed (AAN) paradigm. The controller employs Recursive Least Square (RLS) algorithm to identify human arm impedance parameters, with aims to quantify the residual motor capability of the patient. According to the upper-limb impedance of the patient in the direction along movements, the reference trajectory can be generated based on the motion patterns in healthy humans. In addition, the fuzzy logic control strategy is implemented in the direction perpendicular to movements, which is capable of adjusting the robotic assistance level by considering both upper-limb impedance of the patient and the variations of deviation between the reference trajectory and actual trajectory. Furthermore, the proposed subject-adaptive AAN controller was validated in the simulation study, and the results demonstrated the capability of the controller to modulate the assistance level in accordance with the subject's instantaneous requirements. Future works will focus on the implementation of the control scheme with post-stroke patients.