ELBOW ANGLE ESTIMATION FROM EMG SIGNALS BASED ON MONTE CARLO SIMULATION

Abstract

Monte Carlo simulation is defined as statistical sampling techniques which is used to estimate the solutions of quantitative problems. The aim of this study is to develop Monte Carlo algorithm for elbow angle estimation from EMG signal as preliminary study for further research in rehabilitation tool to make a breakthrough rehabilitation tool for post-stroke patients based on muscle signals to carry out rehabilitation independently and consistently. The Monte Carlo simulation is performed to approach the model’s angle from subject who takes 20 seconds lifting barbell repeatedly for 52 times. Monte Carlo simulations were carried out as many as 10,000 times because it was considered ideal testing for a model. In doing the estimation, the angle will be divided into four ranges, which are determined from the model’s trend value, the estimation of the previous angle, the estimated error angle, and the previous measured angle. Then an average calculation is performed on the Monte Carlo simulation, which enters the angle range to determine the estimated value of the angle. The most optimal estimation is obtained from this study with RMSE (root mean square error) was 8.96°, and the correlation coefficient between estimate angle and the measured angle was 0.96.