Prediction of Driver Intention Using EEG and GSR Signals for Advanced Driver Assistance Systems

As vehicle intelligence increases, driver assistance systems continue to advance to support the driver. Driver condition monitoring is a key component of such systems, which utilize a variety of sensors to collect real-time data about drivers' physiological and psychological metrics. These data are analyzed by computer vision, signal processing, pattern recognition and multi-source information fusion techniques to assess the driver's condition and driving behavior. This study aims to predict the driver's intention to turn left, turn right, accelerate, decelerate, and go straight by fusing EEG and GSR signals prior to the driver's actual maneuvers. The study used Todynet network to obtain a certain prediction accuracy and the validity of the model was verified by simulation environment.