Research On Collaborative Interaction Method Based On AR-BCI

Currently,Augmented reality(AR)-based steady-state visual evoked potential(SSVEP)-based brain-computer interface(BCI)systems are more portable for controlling external devices.However,the classification results of the traditional Electroencephalograph(EEG)algorithm are low when there are many visual stimuli or the stimulation time is short,making it cannot guarantee the real-time performance of the system.In addition,there is a difference between the presentation of visual stimuli in AR and personal computer(PC),and the design of a comfortable AR-SSVEP interaction interface still needs to be investigated.This thesis starts by constructing an efficient SSVEP-BCI fast classification method and exploring whether the color characteristics of visual stimuli affect the classification results of EEG.Furthermore,it verifies the feasibility of the multi-target fast classification method and the effect of stimulus color on the AR-SSVEP classification results through a two-person collaborative operation of the AR-Mind Tic system.The main research work can be summarized as follows:(1)The Convolutional Neural Network(CNN)based multi-target fast classification method for AR-SSVEP was implemented.Similar multi-target visual stimulus layouts were designed in AR-SSVEP and PC-SSVEP,and EEG signals of similar experimental layouts were collected and pretreated.The ability of the CNN model to process data was utilized to achieve efficient EEG classification under short delay.Compared with the classification results of the traditional EEG classification algorithm,the effectiveness of the proposed network model in AR-SSVEP multi-target classification was verified.(2)The effect of visual stimulus color on the classification performance of ARSSVEP was analyzed,which provided a theoretical basis for the selection of visual stimulus color in AR-SSVEP.The four visual stimulus interfaces with different colors were designed,and two kinds of traditional classification algorithms were used to compare the classification difference between AR-SSVEP and PC-SSVEP under different stimulus colors.The signal-to-noise ratio(SNR)and amplitude of EEG signals under different visual stimulus colors were compared.The factors which affect the classification results and EEG signal quality were also analyzed.(3)The AR-Mind Tic collaborative interaction system was built based on the fast classification method of AR-SSVEP and color analysis of visual stimuli.The CNNbased multi-target fast classification method and comfortable visual stimulus colors were applied to the system.Through the real-time shared AR chess interface to realize the collaborative interaction between people,the online experiment results of 5 groups of subjects verified the practicality of the AR-Mind Tic system.