Road safety assessment using Bayesian belief networks and agent-based simulation

Abstract

Road safety performance constitutes an important issue in road traffic management. Systems have been developed for assessing safety performance; however, these provide only historical or retrospective analyses. Effective safety management requires a prospective viewpoint. The main goal of this research is the integration of microscopic road network simulation with Bayesian Belief Network (BBN) technology for improved prediction of road accident risk. The paper describes the method along with the current state of the development of an accident prediction system. Preliminary validation studies of the road network simulation and BBN models are illustrated. .