Effects of Driver Drowsiness on Driving Performance in the Context of Partial Driving Automation Requiring Hands-on-Wheel

Similar to a manual driver, a human using an automated driving system may experience drowsiness. This can be attributed not only to monotonous environments but also to reduced driver activity. Our concern is how can a system identify the driver's state (e.g., states of low arousal and degradation of mental and physical functions) without relying on physiological indices while ensuring vehicle safety. In previous study, we have proposed a dual control theoretic approach, which attempts to simultaneously perform vehicular safety control (prevention of lane departure) as well as identification of the driver's state. The present study applies the proposed approach to partially automated driving systems requiring hands-on driver activity. Sleepy drivers' behaviors and the safety control actions of the applied system were observed and analyzed through the driving simulator experiment. The experimental results show that the proposed system effectively prevents lane departure in the context of partially automated driving.