AUTHOR=Zhang Xiaodong , Lu Zhufeng , Fan Chen , Wang Yachun , Zhang Teng , Li Hanzhe , Tao Qing 
  
TITLE=Compound motion decoding based on sEMG consisting of gestures, wrist angles, and strength
  
JOURNAL=Frontiers in Neurorobotics
  
VOLUME=Volume 16 - 2022
  
YEAR=2022
  
URL=https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/fnbot.2022.979949
  
DOI=10.3389/fnbot.2022.979949
  
ISSN=1662-5218
  
ABSTRACT=To provide a more diversified and flexible operation for the electromyographic hand, this work proposed a demand for upper limb compound motion decoding. In total 60 compound motions were selected, which are combined with 4 gestures, 5 wrist angles, and 3 strength levels. Both deep learning methods and machine learning classifiers were compared to analyze the decoding performance: For deep learning, three structures, and two ways of label encoding were assessed for their training processes and accuracies; For machine learning, twenty-four classifiers, seven features, and a combination of classifier chains were analyzed. Results show, that for this relatively small sample multi-target surface electromyography (sEMG) classification, feature combination (mean absolute value, root mean square, variance, 4th-autoregressive coefficient, wavelength, zero crossings, and slope signal change) with Support Vector Machine (quadric kernel) outstood for its high accuracy, short training process, less computation cost, and well stability (p < 0.05). The decoding result achieved an average test accuracy of 98.42±1.71% with 150 ms sEMG. For separate gestures, wrist angles, and strength levels, the average accuracies were 99.35±0.67%, 99.34±0.88%, and 99.04±1.16%. Among all 60 motions, fifty-eight of them showed a test accuracy as greater than 95% and part equal to 100%.