Research And Implementation Of Brain Computer Interface System Based On Augmented Reality

The technology of Brain Computer Interface(BCI)can connect human Brain,Computer and external devices,realize man-machine interaction and achieve man-machine fusion,instead of relying on traditional peripheral nerves and muscles.Among them,steady-state Visual Evoked Potential(SSVEP)brain-computer interface technology is the most commonly used technology in the field of brain-computer interface,and it has great application value in external device control,rehabilitation treatment and high-risk industries.At present most of brain-computer interface system based on SSVEP shining with multiple fixed block to stimulate target,or improve the stimulus paradigm,the checkerboard,Newton’s rings,as the phase information.Watching stimulation for longtime,it is easy for those participants to cause visual fatigue and pain,even a seizure,and cannot real-time interaction with the real environment.Augmented Reality(AR),on the other hand,is a technology that skillfully integrates virtual information with the real world,allowing users to look at visual stimuli in the same view and interact with the real world.Therefore,this study integrates augmented reality and brain-computer interface technologies to improve user experience and enhance user motivation.The main work of this study was to put forward a combined SSVEP and AR dynamic visual evoked potential(dynamic state visual evoked potential,DSVEP)method,the design experimental paradigm based on the method,using pattern recognition algorithm to analysis user intent,recognition and tracking objects with deep learning algorithms,development of real-time AR-BCI system,implemented using the brain electrical signals to control a humanoid robot.The results provide a deeper understanding of bci and a scientific and effective method for human-computer integration.The main contents of the study include:(1)classification of DSVEP paradigm eeg signalsAccording to the actual environment,the direction of movement of an object in view,size,speed,and so on and so forth design research DSVEP experimental paradigm,the visual stimulus block movement direction,size,speed and scaling of brain electrical signal acquisition,Analysis the impact on the electrical characteristics,and use the Power Spectral Density(PSD),Canonical Correlation Analysis(CCA)and Filter Bank Canonical Correlation Analysis(FBCCA)were used to compare the classification results of EEG signals,and FBCCA algorithm was used to analyze the classification accuracy of static and dynamic stimuli in different time Windows.The results show that the size and direction of the visual stimulus have a weak influence on the EEG signal characteristics,while the speed and speed of scaling are inversely proportional to the signal characteristics,especially to the fundamental frequency energy.The classification accuracy of dynamic stimulus is better than that of static stimulus.(2)object detection research based on deep learningTwo deep learning network models in the field of target recognition were studied.Based on the data set of the research group,two algorithms,YOLOv3 and Faster R＿CNN,were used for comparison experiments.The loss curve and recognition speed of the two models were compared.At last,a better model is selected for the development of real-time AR-BCI system in this study(3)research and development of real-time brain-computer interface system based on ARBuild online brain-computer interface system based on Augmented reality,the overall structure of the design and development system,some key technologies for the analysis of the whole process implementation method,including the BCI system realization of visual interface design,communication interface and an introduction to the robot and control.in addition,this study also used in machine vision for object recognition and tracking,and then the specific frequency of visual stimuli to annotate each object.The multi-threaded design in the system ensures the stable and reliable operation of the online brain-computer interface system.At last,the online experimental results verify that the ar-bci system is stable and reliable,and the object detection and target tracking algorithm is robust.The results of this study show that the ar-bci system designed in this paper is efficient and stable,and the combination of augmented reality and SSVEP is expected to provide a high-performance brain-computer interaction mode,which provides reference value for the development of wearable ar-bci system.