Control of 1-DOF Exoskeleton based on Neural Network Regression Analysis and Wavelet Transform of MES

Improvement of human locomotor performance through external devices such as exoskeletons is a field of active research. This paper presents the design and implementation of an exoskeleton of one degree of freedom (DOF) to assist the flexion and extension of the upper limb. The exoskeleton is controlled by signals from force sensors and myoelectric signals (MES), achieving a reduction of the muscle activity of the user. The MES are captured from the triceps and biceps muscle groups. Subsequent digital signal processing comprises: for feature extraction of signals the time-frequency Wavelet transform is performed, and its following regression analysis is done by an artificial neural network (ANN). We propose a speed control scheme of the exoskeleton from the aforementioned signals, which is executed in real time, achieving a reduction of the biceps muscle activity up to 94%.