Auto-Fusing Covariance and Phase Locking Value With Brain-Inspired Spiking Neural Networks for EEG-Based Driver Reaction Time Prediction

Drowsy driving stands out as one of the major contributors to road collisions. Drowsiness is characterized by a sense of fatigue and a compelling desire to sleep. It manifests through a gradual decrease in reaction time. The electroencephalogram (EEG), which records the patterns of electrical waves in the brain, exhibits a significant correlation with the gradual decline in reaction time induced by drowsiness. This research proposes a superior novel approach that combines phase locking value (PLV) and covariance representations by feature-level fusion by using an autoencoder on brain-inspired reservoir-based spiking neural networks (BI-SNNs) to estimate drivers' reaction times by examining the EEG data. By fusing PLV and covariance features into the reservoir-based BI-SNN method, the network can efficiently capture the spatio-temporal dynamics in the data. The superiority of the proposed methodology is assessed by evaluating the root-mean-squared error (RMSE) and mean absolute error (MAE) on the publicly available lane keeping task (LKT) dataset.