A MODELING OF EXTENDED KALMAN FILTER TO IMPROVE ACCURACY IN ELBOW JOINT ANGLE ESTIMATION

The essential problem in the estimation of a human elbow angle position using myoelectric or electromyography (EMG) is that the EMG features have non-linearity characteristics. The non-linearity of the EMG features influences the performance of the estimation. The objective of this paper is to develop an extended Kalman filter based on the time domain feature to predict the position of the elbow using a myoelectric signal. The contribution of this study is that the non-linearity of EMG feature can be linearized effectively on flexion and extension motion. This is achieved by linearizing the EMG feature in extended Kalman filter using first-order Tailor series. The Ag(AgCl) was used to collect the myoelectric activities from biceps muscle. In this study, the sign slope feature (SSC) extracted the EMG signal to get the evidence that is associated with the position of the elbow. The extended Kalman filter (EKF) was chosen to linearize and to approximate the elbow position using EMG features. The performance of the proposed method is 12.81% and 9.65 % for periodic and arbitrary motion, respectively. We have confirmed the success of the presented EKF method to improve the performance of the estimation. Further, the proposed method can be implemented to an assistive exoskeleton for elderly people or stroke patients for a better life.