Time Delay Analysis of Vehicle Handling Variables for Near-Crash Detection of Drowsy Driving Using a Bus Driving Simulator

Drowsy driving is accompanied by driver’s micro-sleep and overreaction at high levels of drowsiness. During micro-sleeps, drivers fail to respond to changes of the road curvature and hold the steering wheel stationary. After waking up, drivers are shocked and suddenly apply a corrective steering wheel movement. Such behavior is investigated in this research to detect near-crash events caused by drowsy driving. This paper classifies vehicle handling variables in terms of time lag and frequency content. The input is the steering wheel angle and the output signals include the vehicle yaw rate, yaw angle, and lateral position. Modeling, simulation, and experimental results are presented. 15 professional suburban bus drivers perform the tests on a bus simulator. The results show high-order differential handling variables are better indicators of drowsiness compared with low-order handling variables for two reasons: they have lead time and contain a wider range of frequency content. It is shown that the steering wheel angle and the vehicle yaw rate detect the driver’s overreaction faster than the vehicle yaw angle by 1.1 s; and faster than the lateral position by 2.0 s. It is concluded that the steering wheel angle and the yaw rate provide longer lead time to activate safety systems for avoiding or mitigating collisions.