Classification of Driver Distraction Risk Levels: Based on Driver’s Gaze and Secondary Driving Tasks

Driver distraction is one of the significant causes traffic accidents. To improve accuracy accident occurrence prediction under driver and to provide graded warnings, it necessary classify level distraction. Based on naturalistic driving study data, risk levels are classified using driver’s gaze secondary tasks in this paper. The classification results then combined with road environment factors for prediction. Two ways suggested study: divide into three based AttenD algorithm, other six odds ratio. Random Forest, AdaBoost, XGBoost used predict by combining results, characteristics, factors. show that helps model accuracy. better than tasks. method can be applied further warning.