Towards Effortless Brain-Computer Interaction: An SSVEP-Based BCI with Object Detection in FPV

This research addresses one of the key challenges in conventional Steady-State Visually Evoked Potential (SSVEP) Brain-Computer Interfaces (BCIs) used in assistive robotics, specifically, their dependence on mentally fatiguing screen-based flicker stimuli. These traditional systems typically require users to focus on rapidly blinking visual cues displayed on a monitor, which can lead to visual discomfort, cognitive strain and reduced long-term usability. The study aims to enhance the practicality and user-friendliness of BCIs in real-world assistive applications. We introduce a novel SSVEP approach that superimposes flick-ering masks onto detected objects in a real-time first-person view (FPV) video stream captured by a robot. The flickering of these objects is designed to visually stimulate the user and elicit steady-state visual evoked potentials (SSVEPs). Two configurations were evaluated with 11 participants: a fixed-position SSVEP-control panel for navigation and an object-specific control using YOLOv5 for dynamic object selection. Compared to a benchmark LED-based SSVEP dataset, our FPV methods achieved competitive average classification accuracies of 90% for the control panel and 91% for object-based interaction, despite lower signal-to-noise ratios due to video background noise. These results demonstrate the feasibility of an intuitive BCI-robot interface that shifts complexity to computational modules while maintaining end-user ease of use.