A series of driving simulation pilot studies on various technologies for alertness monitoring (head position sensor, eye-gaze system), fitness-for-duty testing (two pupil-based systems), and alertness promotion (in-seat vibration system) has been conducted in Circadian Technologies' Alertness Testbed. The results indicate that all tested technologies show promise for monitoring/testing or preventing driver fatigue, respectively. However, particularly for fatigue monitoring, no single measure alone may be sensitive and reliable enough to quantify driver fatigue. Since alertness is a complex phenomenon, a multi-parametric approach needs to be used. Such a multi-sensor approach imposes challenges for online data interpretation. A neural-fuzzy hybrid system is suggested for the automatic assessment of complex data streams for driver fatigue. The final system output can then be used to trigger the activation of alertness countermeasures.
Abstract