Impairment of driving performance due to drowsiness or fatigue is a major cause of severe traffic accidents, which is motivation for this work on driver monitoring, especially drowsiness detection. To include the physiological correlations of drowsiness in car driving, an extensive experimental study was conducted at the corporate research centre of Robert Bosch GmbH.
During this work, a lot of efforts have been invested in the design and test of the driving simulator. The following study with a cohort of 66 subjects lead to 44 reliable data sets in which the main objective of sleep induction was obtained in most of the trials.
An extensive literature review refines the methods and establishes the basis for the verification of the recorded data and the applied methods by using already approved approaches. Extending this state of the art, a novel approach is described, which is driven by consideration of physiological changes during drowsiness with the intention of predicting dangerous level of drowsiness by analyzing driving behavior. As already known, the measurement of physiological signals is possible in laboratory conditions for setting up a drowsiness reference, but in everyday environment its usage is either too complicated or the signals can not be recorded in the desired quality.