Research Of EMG-based Control Methods For Manipulator

The control process of robotic arm based on surface Electromyography signals(sEMG)is actually analyzing and recognizing the forearm sEMG signals by machine,and then control the robotic arm to make the appropriate action.This technology can help the amputees control the mechanical arm through their own intention,thereby reducing motor dysfunction and improving their quality of life.At present,the research of control method based on sEMG signals at home and abroad obviously has shown the characteristics of low accuracy and stability.In order to achieve the accurate recognition of hand movements and the real-time control of the manipulator,based upon a detailed analysis of the present worldwide research status,the study of EMG-based control methods based on pattern recognition was focused on in this paper.The research content of this paper was divided into the following aspects:(1)The generation mechanism and characteristics of the sEMG signals were introduced in this acticle,and the position of electrodes was determined according to the function of human forearm muscles.In addition,a two-channel acquisition platform of sEMG signals was built,which can collect the signals according to different sampling frequency.At the same time,wavelet threshold denoising algorithm was used to denoise the original signal in order to ensure the quality of signals.(2)For extracting more effective information,the sEMG signal was analyzed from three aspects: time domain,frequency domain and time-frequency domain.The mean absolute value(MAV),root mean square(RMS)and modulus maximum of wavelet coefficients were used to construct the feature vector after a detailed comparison.Moreover,the laplacian eigenmaps(LE)algorithm was used to reduce the dimensionality of the high-dimensional feature.Then the reduced dimension feature vector was used as the input of classifier,which would be helpful to the pattern recognition of different gestures.(3)According to the basic principle of pattern recognition methods used in the field of EMG signal widely,the advantages and disadvantages of them were analyzed in detail.On this basis,the VPMCD algorithm was introduced into the field of EMG signal.Further more,VPMCD algorithm,BP neural network algorithm and SVM algorithm were compared in this paper.The results showed that VPMCD algorithm could classify the human forearm EMG signal more effectively,which has better recognition performance.(4)The EMG control platform of the robotic arm was developed based on the LabVIEW software.The whole platform was divided into the following modules: EMG signal acquisition module,pattern recognition module and communication control module.Finally,the online myoelectric control experiment based on the VPMCD recognition methods was made,and the ability of grasping objects by the mechanical arm was realized in this study.