Driver Distraction Assessment Using Driver Modeling

Characterizing individual human drivers is of increasing interest for applications like adaptive driver assistance or monitoring. Describing the human driver by means of control-theoretic driver models constitutes a promising approach. In this paper, we apply a driver model adopted from literature to real-road driving of a distraction experiment in order to assess the driver state. The control-theoretic driver model features an anticipatory and a compensatory tracking component as well as a processing delay and a neuromuscular motor component. The distraction experiment data comprises real road driving with a visuomotor and an auditory secondary task, as well as reference driving. By means of prediction error identification, we continuously and individually estimate the model parameters from driving data of eleven drivers. We evaluate the distributions of the driver model parameters and the predictive capability of the estimated driver models. The estimated driver model parameters reflect distracted driving behavior according to the driving task. As a promising experimental result, the driver model parameters and predictive performance are significantly associated with driver distraction.