Research On Robot Control Method Based On The Technology Of Hybrid Brain-Computer Interface

With the continuous development of brain-computer interface technology,the brain-computer interface system of single paradigm can not meet the requirements of the robot to complete high-complexity tasks because of the fewer control commands.As a result,the hybrid brain-computer interface has become a research mainstream.In this thesis,a robot control method based on hybrid brain computer interface is designed.The task of robot control is realized by detecting alpha-block,steady-state visual evoked potential(SSVEP)and P300 signals.The research work of the thesis mainly includes the following:First of all,the robot control system based on hybrid brain computer interface is established.Based on the analysis of robot kinematics,the virtual two degree of freedom planar manipulator is built by using OpenGL as the controlled object of brain-computer interface.Considering the factors of stimulus frequency,stimulus interval,background color and so on,the stimulation inducing interface of SSVEP and P300 was designed under the environment of VS2010,and the acquisition and processing system of EEG signal was established.Then,feature extraction and classification of alpha-block,SSVEP and P300 signals are completed.The alpha-block identification is realized by using the amplitude-frequency characteristics of fast Fourier transform for the switching of two visual stimulation interfaces.The power spectrum and canonical correlation analysis are used to detect SSVEP and determine the frequency of subjects’ gaze.In the aspect of feature extraction of P300 signal,in order to reduce the data dimension and training time,the thesis combines the time-domain energy entropy and low-frequency wavelet decomposition coefficient as its features.In view of the low accuracy of the hybrid brain computer interface system,a voting decision method based on linear discriminant analysis(LDA),support vector machine(SVM)and time-domain analysis is proposed to recognize the P300 signal.Finally,online experimental verification and result analysis are carried out.On the basis of offline analysis,ten subjects participated in the robot control experiment based on hybrid brain-computer interface.Through the analysis of performance index,the average accuracy of the three signals is 90.87%and the experiment success rate is over 94%.Subjects can freely control the movement of the manipulator to the target,which verified the superiority and reliability of the system.The robot control system based on the hybrid brain-computer interface technology proposed in the thesis enriches the control commands of the robotic arm.The dimension of P300’s feature vector is reduced by time-domain energy entropy and wavelet decomposition.Compared with classification of LDA or SVM alone,the method of the voting decision improves the accuracy of the system.At the same time,the feasibility of the new paradigm that combines alpha-block,SSVEP and P300 signals to control robot system is verified.