Abstract:
Currently there are many people who have walking problems. This research aims to develop and solve these problems by introducing walking assistance system which can recognize 3 types of gestures, that include walking, sitting and standing. Our system is divided into 3 main parts including classification, posture training for the exoskeleton and exoskeleton suit systems. Conjugate Gradient Backpropagation Neural Networks have been used to classify sEMG signals of lower limb postures after extracting the features. Then the output of classification has been used to command the exoskeleton suit to perform the gesture according to the results of the recognition. This research aims to present 2 methods for recognizing the suit gesture: fixed programmed gestures and training gestures. In addition, our work uses PID controller to control the DC motors. In order to reduce the number of motors and increase stability of the system, the of Four Bar Linkages Mechanisms of Lower Limb Exoskeleton suit are used. The results from the training gesture method have shown that the Levenberg-Marquardt method provides the lowest average error of 4.418, while the general method of curve fitting gives 12.71. The results from classify sEMG signals of lower limb postures show that all features in time domain provided the highest recognition rate at 99.39%.