Synergy Estimation for Myoelectric Control Using Regularized NMF

This paper investigates the problem of achieving robust control for assistive devices using surface electromyogram. We propose the use of regularized non-negative factorization using alternating least square algorithm for the estimation of muscle synergies. The proposed algorithm is tested for 3 different degrees of freedom only one active at a time, providing 97.6 - 99.5% accuracy and when compared with previously proposed methodology it reduces the delay time by 45%.