The detection of drowsy drivers through driver performance indicators

Driving while drowsy is a significant contributor to motor vehicle accident deaths as drowsiness impairs a driver's reaction time and control over the vehicle. The primary goal of this research was to determine which, if any, driving performance variables are closely related with empirical measures of driver drowsiness. This effort was undertaken in an attempt to create the foundations for an innovative drowsy driver detection system based on driver performance metrics alone, or that may be integrated with other technological approaches. This research is the first known attempt at linking data from a closed-track study using electroencephalogram (EEG) with driver performance measures to examine the transition from an alert to drowsy driver state.;Nine sleep-restricted participants drove on a closed test track for a two-hour period during the late night or early morning hours. EEG data and various channels of driver performance data were collected continuously throughout the drive. These two datasets were then synchronized and compared to examine changes in driver performance as drivers transitioned from being alert to drowsy, and eventually, in most cases, falling asleep at the wheel. Video data of the driver and the forward view of the vehicle were also collected and used to code and perform an analysis of the drivers' behavior, validate the EEG data, and analyze differences in behavior among subjects.;While no statistically significant relationships between driver drowsiness state and driver performance were found, this research is an important first step toward addressing this complicated problem, and provides insight and direction for future research. While additional research should be done to make any specific claims, this research suggests that; due to large individual differences between drivers a within-subjects approach to using driver performance metrics to detect driver state may be more realistic approach than using a universal detection method, steering range and steering variability were found to have the strongest relationship to driver drowsiness and would be good variables to investigate in future drowsy driver research. Also, neither eye closures nor visually scored EEG are sufficient measures for detecting drowsiness in drivers.