Abstract

Human force estimates have a wide range of applications, including biomedical models, rehabilitation, biomechanical system control, and human-machine interfaces. The development of methods for efficiently estimating force to enable such applications is necessary and challenging. In this paper, we propose a system for force estimation using eight-channel surface electromyography (sEMG) signals measured from human arm muscles. The sEMG signals are collected from the Myo armband, while the force signals are gathered from a mobile robot handle for rehabilitation applications. The novelty and contribution of this work is that the real-time force can be estimated using a novel weighted optimization method, as our proposed solution, where the optimal weights of all sEMG channels and the optimal offset value are automatically calculated and used for force estimation. Experimental results demonstrate that the proposed method can efficiently estimate forces in four directions: X +, X-, Y +, and Y-. The estimation errors measured by an average root mean square error (RMSE) are between 1.581 and 1.882 N. The computational cost analysis of the proposed method in terms of mathematical operations is also discussed. Finally, our results are compared with the results reported in the literature.