Computational Modeling of Driver Distraction by Integrating Cognitive and Agent-Based Traffic Simulation Models

The U.S. National Highway Traffic Safety Administration has reported that the major cause of vehicle crashes in the United States has been driver inattention and distraction. Among various types of distraction, cell-phone dialing and text messaging have been shown to significantly degrade driving performance. Existing studies have mainly focused on driver distraction scenarios with one or a few vehicles, and have generally not looked at large-scale simulation of many vehicles. In contrast, currently available software for simulating traffic with many vehicles (e.g., CORSIM and VISSIM) has generally not considered the effects of driver distraction on driver/vehicle behavior. To overcome these limitations, we propose a computational modeling framework that integrates a cognitive model of distraction (i.e., Distract-R) and an agent-based traffic (micro-) simulation model (i.e., VISSIM). The framework aims to enable transportation modelers to easily set up virtual experiments and evaluate the impact of distracted drivers on large-scale networks. We validate the framework using two existing experimental data sets. Preliminary results indicate that the framework can account for statistically significant changes in speed fluctuation and headway distance in the presence of a significant number of distracted drivers.