Abstract
Muscle activation analysis is prominent for finger kinematics analysis which opens the scope in various domains including prosthetic palm. In this paper, a synergetic analysis of muscle activation is done with finger kinematics. The surface electromyogram signals are recorded from extrinsic muscles of the hand with various types of finger flexion–extension. The linear regression (LR) and random forest-based nonlinear regression analysis are performed for model development and comparative evaluation. The LR model is used to capture the linear relationship between surfaces electromyography (sEMG) signals and finger kinematics. However, random forest regressor (RFR) is employed to model the nonlinear relationship between sEMG signals and finger kinematics. Subject-specific and type-of-finger flexion action-specific analysis claimed that the RFR model achieved root mean square error of 0.487 which shows that the average deviation between predicted and actual values is very less. The R2R2 score 0.99 is observed using RFR indicating that the model can successfully explain 99% of the variance of the data. Finally, the average testing time of 0.0274 ms is observed for the RFR model. The superiority of the developed model is justified in comparison with the other developed models. The findings of this study not only present a viable approach for accurate and intuitive myoelectric control, which uses muscle-generated electrical signals for prosthetic device operation but also provides a new perspective on muscle synergies.
