Utilizing Error-Related Potentials to Assist in Decision-Making at Crossroads

This study initially conducted an offline simulated driving experiment to detect Error-Related Potentials (ErrPs) produced by subject s, achieving an average classification accuracy of 79.99% using the Polynomial Kernel Support Vector Machine (PSVM) classification algorithm. Subsequently, an online experiment on human-machine cooperative driving was carried out to assess the feasibility of applying ErrPs in real-time cooperative driving scenarios. The results indicated that while rule-based autonomous driving systems ensured basic safety, the integration of ErrPs into human-machine cooperative driving further incorporated the drivers' decision preferences. This research validates the practical feasibility of utilizing Error-Related Potentials for human-machine cooperative driving and offers a new direction for the development of EEG-assisted human-machine cooperative driving systems, with the aim of enhancing system performance and driving experience.