Simplified, data-driven, errorable car-following model to predict the safety effects of distracted driving

Abstract

An errorable car-following model is presented in this paper. The model was developed to predict the situational risk associated with distracted driving. To obtain longitudinal driving patterns, this paper analyzed and synthesized the NGSIM naturalistic driver and traffic database to identify essential driver behavior and characteristics. NGSIM data was modified according to data from cognitive psychology concepts to examine the probabilistic nature of distracted driving due to internal vehicle distractions. The errorable microscopic car-following model was developed and validated, which can be fully integrated with the naturalistic data and incorporate the probabilities of driver distraction. The proposed model predicts that distracted driving in congested conditions can result in crash rates 3.25 times that of normal driving conditions.