The purpose of designing and manufacturing prosthetic organs is to create their maximum behavioral similarity to human organs. The aim of this paper is to improve the robotic arm control via Model Reference Adaptive System (MRAS) based on Lyapunov theory using EMG data classification. In this paper, human arm is modeled with a robot with two degrees of freedom. The proposed control method is MRAS. The outcome of this research is a robotic arm with MRAS, using the classification of electromyogram (EMG) data recorded from human arm movements, results in proper tracking of the reference signal, less overshoot and steady-state error compared to the conventional PI controller. For this purpose, using two electrodes, EMG data is collected from the anterior deltoid and middle deltoid muscles of the arm of five female athletes and by performing two movements of abduction and flexion of the arm. Then, after eliminating noise, integral of absolute value (IAV), zero crossing (ZC), variance (VAR) and median frequency (MF) are extracted. Then, classification is done by linear discriminant analysis (LDA) method to detect movements based on data characteristics. Finally, the proposed controller and model are designed according to the EMG characteristics to achieve the proper control response and the appropriate command signal is sent to the controller to perform the corresponding movement. The results and the values of the obtained errors show the conformity of the model and controller behavior with the predefined movement pattern.
Manuscript profile
