Bio-signal interface for knee rehabilitation robot utilizing EMG signals of thigh muscles

Abstract

A sitting knee rehabilitation robot has been developed in order to work with biological signals of patients muscles (EMG). This type of robot is capable to perform all active and passive exercises to improve the functionality of injured knee and thigh muscles. Intention of patients/operators to move their knee is detected by electrical activities of knee muscles (EMG) recorded above an observed muscle. A proportional support is then applied to patient's knee by means of the robot arm. In order to estimate the torque generated in the knee joint simple biomechanical model of knee joint have been used. Some parameters of this model represent individual biomechanical properties of the human body differ from patient to patient. Therefore, a calibration program has been performed to determine these parameters individually. To estimate real torque produced in knee joint calibrated parameters are inserted into the biomechanical model. An orthopedic brace is designed to measure the knee forces in both flexion and extension position of knee throughout calibration process. Experiments on a healthy subject have indicated that the interface is able to estimate the actual knee torque and also capable to support the patient's knee during movement by providing part of the predicted torque.