Analysis of Elbow Joint Angle for Prediction based on EMG using Kalman Filtering Technique

Abstract

In this study, an accurate estimation of elbow joint angle by measuring surface electromyography(sEMG) signal from human biceps muscle is done using machine learning regression methods such as linear and polynomial regression. The result of the regression technique is further validated using Kalman Filter Technique (KL) which gave better accurate results for the angle taken. As the first step of regression analysis, the sEMG signal measured from the human biceps muscle, using the Myoware IC has been preprocessed to make it suitable for further analysis. Then the second step was to extract the features from the measured sEMG signal and in this paper, four features of the time-domain method were extracted to estimate the elbow joint angle, namely integrated EMG (iEMG), LOG, RMS and Mean. The extracted features were then applied to the machine learning regression algorithm to predict the elbow joint angle. The predicted elbow joint angle using regression and the Kalman filter showed that the results found using the Kalman filter gave higher accuracy than polynomial regression.