Controlling an Upper-Limb Rehabilitation Robot by EMG Signals

This paper presents a method for controlling the motion of a 6-DOF upper limb rehabilitation robot based on electromyography (EMG) signal. A kinematic model has been determined by Denavit - Hartenberg (D-H) method. Due to the complexity of the robot mechanical structure, the inverse kinematics is addressed using the Jacobian method. Besides, the EMG signal is a physiological signal generated during muscle contraction, which is collected from a low-cost sensor interacted on the human arm, but noise is inevitable in the extracted data. Therefore, the method treats EMG signal noise using Butterworth filter and proposes a method to predict elbow joint angle based on EMG using zero crossing and Kalman filter. Finally, the proposed method is evaluated through simulation on MATLAB Simulink and experiments on a 6-DOF robot arm model. The experimental results show that the EMG signal processing method proposed is significantly effective and the upper limb rehabilitation robot based on EMG signals is feasible.