A linear model for simultaneously and proportionally estimating wrist kinematics from emg during mirrored bilateral movements

This paper presents a linear model for simultaneous and proportional estimation of the two degree-of-freedoms (DOFs) wrist angle positions with surface electromyography (EMG). A 5th order state-space model was used to estimate wrist kinematics from 4-channel surface EMG signals of the contralateral forearm during mirrored bilateral movements without motion constraints. The EMG signal from each of the three limbed normal subjects was collected along with each angle position in two DOFs from both of the arms, with motion parameters tested including the radial/ulnar deviation and flexion/extension of the wrist. The estimation performance was in the range 0.787-0.885 (R² index) for the two DOFs in three limbed normal subjects. The results show that wrist kinematics can be estimated in 2 DOFs by state-space models with relative high accuracy compared with the results reported previously. The method proposed, as requiring only kinematics measured from the contralateral wrist, is potentially available for a unilateral amputee in simultaneous and proportional control of DOFs in powered upper limb prostheses.