Evaluating the performance of Kalman filter on elbow joint angle prediction based on electromyography

High accuracy in joint angle prediction is very important in the development of devices based on myoelectric control. Joint angle prediction based on a non-pattern recognition method is more preferred due to the robustness and the easiness to adapt to any subject. This paper proposed a new method to predict an elbow joint angle based on electromyography (EMG) which used a time domain feature, zero crossing, and Kalman filter. The EMG signals were collected from biceps muscle using Ag (AgCl) electrodes. To test the proposed method, the subjects were asked to move the elbow in the flexion and extension motion at different periods of motion (12 seconds, 8 seconds and 6 seconds). The EMG features yielded from the feature extraction step were processed using a Kalman filter which was used to predict an elbow joint angle. The performance of the proposed method was evaluated using root mean squared error (RMSE) and the Pearson’s correlation coefficient (CC). In this study, the RMSE and CC values were ranged from 6.9o to 17.5o and 0.93 to 0.99 respectively. The results of the experiment have demonstrated the effectiveness of the proposed method to predict an elbow joint angle based on EMG signal.