Real-Time EMG Signal Processing with Implementation of PID Control for Upper-Limb Prosthesis

In this research feedback control of prosthetic arm has been presented. It aids people, those who are suffering from trans-humeral amputation based on Electromyography (EMG) signals. The acquired signals are utilized to generate a control command for the movements of elbow joint. These would compensate for the lost function of flexion-extension motion from the ulna-humeral joint and wrist pronation-supination from the proximal radio-ulnar articulation of the forearm. Myo Armband is used to acquire EMG signal from biceps and triceps muscles. By integrating the simulation with realtime measurement of EMG signals from selected muscles helped in data acquisition and classification for desired motions. Offline as well as online classification accuracy was evaluated by experimentation including ten ablebodied participants. The offline training is done using Artificial Neural Networks with an accuracy of 94%. For real-time analysis, Support Vector Machine is used for classification with an accuracy of 85%. For high-speed processing and portable system, Raspberry Pi is used as it provides high functionality. Further, five control commands are generated to control device movements which include elbow extension and flexion, wrist pronation and supination along with rest state. Feedback control of 2 degree of freedom prosthetic Arm is simulated and implemented using PID control algorithm. The simulations were carried out in MATLAB
^®
as it provides smooth control by reducing the error and making sure that the prosthetic device reaches its desired position.