Research on signal processing of upper limbs of three gestures

Abstract

There are a large number of disabled people in the world whose lives are seriously affected by the lack of upper limbs. Research on related prostheses is crucial to making these individuals’ lives as convenient as those of regular people. The association between the acts that took place and the relevant data waveforms are determined in this investigation. To lessen the impact of fluctuations in the data itself, EMG signals are obtained using the five channels’ worth of data. By examining the related EMG data, preprocessing, feature extraction, and the construction of two CNN models are used to categories three gestures (scissors, rock, and cloth). Finally, the classification accuracy and loss indicate that the 2-D CNN model is better by comparison with the 1-D CNN model in 3 gesture classification. These correlation results show that the control of the prosthetic limb can realize the completion of various actions through the analysis of the corresponding data and transmission to the entity.