Evaluating the linear regression of Kalman filter model on elbow joint angle estimation using electromyography signal

Abstract

High performance in the elbow joint angle estimation, using electromyography (EMG) signal, is essential in the devices which uses EMG as a control signal. The purpose of this study is to evaluate Kalman filter performance which is used to improve the linearity of the elbow joint angle estimation. The EMG signal is recorded from biceps muscle while the elbow performed a flexion and extension motion with two second periods of motion. In order to acquire the information which is associated with the elbow joint angle, the EMG signal is extracted for every 100 milliseconds by using the twelve time-domain features. Kalman filter is used to improve the elbow joint angle estimation after feature extraction process. The findings show that the maximum performance of the elbow joint angle estimation is 0.96±0.01 (correlation), 0.93±0.02 (R2), 9.02°±6.82° (intercept), 0.89±0.09 (slope) from sign slope change (SSC) feature. This proposed method has revealed the effectiveness of Kalman filter to improve the performance of the elbow joint angle estimation based on EMG signal.