Enhanced System Robustness of Asynchronous BCI in Augmented Reality Using Steady-State Motion Visual Evoked Potential

Area of Science:

• Neuroscience
• Human-Computer Interaction
• Biomedical Engineering

Background:

• Augmented reality (AR) headsets offer immersive potential for brain-computer interfaces (BCI).
• Evaluating BCI performance under varying background conditions is crucial for practical AR applications.
• Steady state visually evoked potentials (SSVEP) and steady state motion visually evoked potentials (SSMVEP) are common BCI paradigms.

Purpose of the Study:

• To assess the impact of background changes on SSVEP and SSMVEP BCIs within an AR headset.
• To compare the signal robustness and classification performance of SSVEP and SSMVEP in AR environments.
• To evaluate advanced decoding methods for BCI accuracy in AR.

Main Methods:

• Implementation of a four-target SSVEP and SSMVEP BCI using the Cognixion AR headset prototype.
• Comparison of active (AB) and non-active background (NB) conditions.
• Offline analysis utilizing canonical correlation analysis (CCA) and complex-spectrum based convolutional neural network (C-CNN).
• Evaluation of asynchronous pseudo-online performance for the SSMVEP BCI.

Main Results:

• SSMVEP stimuli demonstrated superior robustness to background variations compared to SSVEP in the AR setting.
• The C-CNN method outperformed CCA in decoding accuracy for both SSVEP and SSMVEP, particularly with non-active backgrounds.
• Offline accuracies using C-CNN (W=1s) were: SSVEP (NB: 82%±15%, AB: 60%±21%) and SSMVEP (NB: 71.4%±22%, AB: 63.5%±18%).
• For W=2s, AR-SSMVEP BCI with C-CNN achieved 83.3%±27% (NB) and 74.1%±22% (AB).

Conclusions:

• AR-SSMVEP BCIs, when combined with the C-CNN method, exhibit significant robustness to background changes and high decoding accuracy.
• The findings suggest SSMVEP BCIs are a promising and practical choice for AR applications, outperforming SSVEP BCIs in challenging visual environments.
• This research highlights the potential of low-cost AR headsets for developing effective SSMVEP-based BCIs.