Performance Research And Application Of Steady-State Visual Evoked Potential Brain-Computer Interface Based On Augmented Reality

The biggest advantage of the Brain-Computer Interface(BCI)based on SteadyState Visual Evoked Potentials(SSVEP)is its large instruction set and high Information Transfer Rate(ITR).Almost all current SSVEP-BCIs use a computer screen(CS)to present visual stimuli,which limits its flexible application in practical scenarios.Augmented Reality(AR)technology provides the ability to superimpose and fuse the stimulus interface of SSVEP-BCI with the real world,which greatly expands the usage scenarios of SSVEP-BCI.But whether we can maintain the performance advantage of SSVEP-BCI after we transfer visual stimuli to AR is a question.In this paper,we first compared and analyzed the performance of AR-SSVEP and CS-SSVEP under different stimulus paradigms,then studied the generation mechanism of AR-SSVEP from the perspective of dynamic brain network,and finally verified the performance of ARSSVEP in practical application scenarios.(1)We designed four stimuli paradigms with different numbers of stimuli and presented them in AR and CS,respectively.Three commonly used recognition algorithms were used to analyze the effects of different stimulus paradigms on the performance of AR-SSVEP and compared with CS-SSVEP.The results showed that the amplitude and signal-to-noise ratio of AR-SSVEP were different from CS-SSVEP.Further comparative analysis found that in AR-SSVEP,SSVEP recognition accuracy decreased as the number of stimuli increased,but not in CS-SSVEP.When the number of stimuli increased,the ITR of AR-SSVEP first increased and then decreased,while the maximal ITR of CS-SSVEP increased all the time.(2)There is a big difference in performance between CS-SSVEP and AR-SSVEP,and we hope to explain this difference from the mechanism.We investigated the brain network topology of SSVEP induced in AR and CS environments,respectively.The results showed that AR-SSVEP formed a relatively stable network structure and a central node located in the occipital region at 800ms.CS-SSVEP had a stable network structure at 500ms and a central node located in the parietal lobe.And the central node degree of CS-SSVEP was higher than that of AR-SSVEP.The above results suggested that the speed and topology of effective brain network formation may be the main factors affecting the performance of AR-SSVEP when different stimulus presentations induce SSVEP.(3)In this study,an augmented reality-based SSVEP-BCI car control system was built.The system is mainly programmed using a mixture of C#and MATLAB,using the Holo lens generation to present a stimulus paradigm.Referring to the previous research,we chose FBCCA,which did not require training and had better performance,as the recognition algorithm of the system.The function algorithm module of the system outputed commands,and wirelessly transmitted commands to control the movement of the trolley.