A hybrid adaptive multi sensor data fusion for estimation of skeletal muscle force for prosthetic hand control

Effective use of upper extremity prostheses depends on the two critical aspects of precise position and force control Surface electromyographic (sEMG) signals can be used as a control input for the position and force actions related to the prosthesis. In this paper, we use the measured sEMG signals to estimate skeletal muscle force. Further, we consider skeletal muscle as a system and System Identification (SI) is used to model multi-sensor sEMG and skeletal muscle force. The sEMG signals are filtered utilizing optimized nonlinear Half-Gaussian Bayesian filter, and a Chebyshev type-II filter provides the muscle force signal The filter optimization is accomplished using a Genetic Algorithm (GA). Multi-linear and nonlinear models are obtained with sEMG data as input and skeletal muscle force of a healthy human hand as an output for three sensors. The outputs of these models for three sensors are fused with a probabilistic Kullback Information Criterion (KIC) for model selection and an adaptive probability of KIC. The final fusion based force for multi-sensor sEMG gives improved estimate of the skeletal muscle force.