Enhanced Brain-Controlled Mobile Robot Based on SE-VEP Paradigm With Single Stimulus

Brain-computer interface (BCI) systems based on steady-state visually evoked potentials (SSVEPs) have been widely adopted because of their efficiency and accuracy. However, the traditional SSVEP method has limitations, including visual fatigue and interference between different stimuli. To address these issues, a new BCI paradigm, namely, a spatial encoding-visually evoked potential (SE-VEP) model, is proposed in this work. This paradigm involves deploying four target points to implement gaze restrictions around a stimulus block and optimizing the locations of these target points through offline data acquisition. This design facilitates electroencephalogram (EEG) encoding for four instructions while using only one stimulus block. Data with varying eccentricities are classified using the Riemann kernel-based support vector machine (R-SVM) approach, which achieves a classification accuracy of up to 86.11%. As the eccentricity increases, the classification accuracy initially increases but subsequently decreases. By evaluating the information transfer rate (ITR), the optimal time window length for online BCIs is determined to be 1.2 s. Additionally, an online brain-controlled robotic virtual system is developed to validate the feasibility of the proposed paradigm for online brain-computer interface applications. The results confirm the effectiveness of the proposed paradigm in implementing an online BCI control system. An evaluation conducted with scales and the information transfer rate for a single stimulus (ITRSS) indicates that compared with the traditional BCI system, the proposed paradigm yields greater reductions in user fatigue (2.8 ± 0.5 vs. 4.1 ± 0.6) and stimulus block utilization (24.6 ± 2.3 vs. 8.2 ± 1.1 bits/min).