A robust method to detect driver distraction
Gert Weller, B Schlag · 2010
Distraction is regarded as the sole or most important contributing factor to accidents. Therefore, methods are needed which allow the reliable detection of distracted driving. For the results reported here, were collected during on the road driving experiments with an equipped vehicle. Besides driving behaviour, gaze was recorded during distracted and non-distracted periods of driving. The were first analysed for differences between both conditions. Based on these results the variable standard deviation of the gaze data was selected for further analyses. By performing a binary logistic regression it could be shown that this variable could reliably predict states of distraction. With the technical progress in gaze detection these results are regarded as highly promising for future solutions of in-car detection of driver distraction.