Towards a countermeasure device to detect fatigue in drivers

Abstract

Fatigue affects the drivers' ability to continue driving safely. Therefore, on-line monitoring of physiological signals while driving provides the possibility of detecting fatigue in real time. The EEG signal has been found to be the most predictive and reliable indicator. However, little evidence exists on implementing EEG into a fatigue countermeasure device. The aims were to utilise EEG changes during fatigue for development of fatigue countermeasure software and to test the ability of such software in detecting fatigue. EEG was obtained in twenty truck drivers during a driver simulator task till subjects fatigued. Changes found in delta, theta, alpha and beta activity were used to develop algorithms for the software. The software was designed to detect an alert state and early, medium and extreme levels of fatigue. The software was tested in off-line mode in a group often truck drivers. The software was capable of detecting fatigue accurately in all ten subjects. The percentage of time the subjects were detected to be in a fatigue state was significantly different to the alert phase (p<O.O1). For 40% of the total driving time subjects were alert and for 60% of the time, the software detected one of the three fatigue states. In on-line analysis the software could alert the three stages of fatigue. The software could detect fatigue accurately. This is the first countermeasure software that can detect fatigue based on EEG changes in all bands. Future field research is required with the fatigue software to produce a robust and reliable fatigue countermeasure system.