An Enhancement of the Driver Distraction Detection and Evaluation Method Based on Computational Intelligence Algorithms

Abstract

Driver distraction is a fundamental problem for human safety, because the number of traffic accidents due to distracted driving does not decrease. In this paper, an enhancement of previously proposed driver distraction detection and evaluation methodology is introduced. The method is composed of computational intelligence algorithms: a driver performance prediction algorithm with nearest neighbor regression and an intelligent fuzzy logic evaluation algorithm. Thanks to the improvement, an additional variable for driver performance prediction and an additional performance-based indicator were introduced. To verify the novelty, the series of thirty driver-in-the-loop experiments has been delivered on an industrial vehicle simulator. At this, an interaction with a vehicle on-board computer was exploited as a distractive activity. Finally, the enhanced method is compared to the previously described one.