Simultaneous and Proportional Estimation of Finger Joint Angles from Surface EMG Signals during Mirrored Bilateral Movements

摘要

This paper studies the continuous estimation of finger angles with state-space model to realize the simultaneous and proportional control of finger joint motions, where the test model of mirrored bilateral movements is first applied to decode the fingers’ continuous motion. First a linear method was proposed for simultaneous and proportional estimation of 10 degrees of freedom (DOFs) finger joint angle with surface eletromyography (EMG). The state-space model was adopted to estimate one finger joint angle from 8 channel surface EMG signals recorded from contralateral limb mirrored bilateral movements. The average estimation performance of 8 DOFs of the hand except little finger can achieve 0.824(index R2), while the index R2 can achieve as high as 0.924 with the estimation thumb metacarpophalangeal joint (MCP). Although the estimation performance of the little finger was not good enough, the results demonstrate that state-space model is a feasible method to estimate the finger joint angles. Because the proposed method requires only the contralateral finger joint angles as the model identification reference, there is a great potential application for unilateral amputees in control of dexterous myoelectric prosthesis with multiple DOFs.