In general, it is difficult to recognize EMG patterns for controlling prosthetic arms because they are not fixed and vary greatly in characteristics. In this paper, we evaluate the classification ability of root mean square, integral, and power spectral densities to identify various hand motions including wrist flexion, extension, wrist adduction, abduction, grasping, and abduction. To perform this study, EMG signals are measured in the finger extensors, carpal tunnel extensors, carpal tunnel flexors, and carpal tunnel flexors. EMG signals are analyzed for root mean square, integral and power spectral density. Through experiments, it was shown that the proposed method can estimate hand movements efficiently.
Keywords
EMG signals, EMG patterns, finger extensors, hand motions and prosthetic arms.