Incremental Adaptive Gesture Classifier for Upper Limb Prostheses

Myoelectric pattern recognition is widely used to control upper limb prostheses. However, the non-stationary characteristics of electromyography (EMG) signals, caused by physiological changes (e.g. muscle fatigue) or non-physiological electrode- skin impedance), hinder use prostheses and deteriorate performance myoelectric system. In this paper, a set robust features proposed be integrated with adaptive learning techniques in order improve performance. Four types are proposed, namely H, HI, AI, AIH features. The H correspond histogram-oriented gradient (HOG) algorithm High-Density surface Electromyography map (HD-sEMG map). On other hand, HI generated combining intensity feature that evaluated from HD-sEMG map. AI calculated segmented maps constructed individual channel. Finally, obtained Offline online tests conducted evaluate results show employing an classifier improves classification accuracy 91.58% 97.2% setups. Results also more against noise than TD average reduced 0.7% when additive White Gaussian applied, comparison 5.8% reduction used. confirm robustness extracted