Multiregime Sequential Risk-Taking Model of Car-Following Behavior

Abstract

Car-following models constitute the main component of operational microscopic simulation models and are intended to capture intervehicle interactions on highway sections. Most existing car-following models are deterministic and do not capture the effects of surrounding traffic conditions on the decision-making process of the driver. An extension to a previously introduced sequential risk-taking model is offered to capture the effects of surrounding conditions on driving behavior. The model extension recognizes two behavioral regimes that depend on the complexity of the decision situation associated with the prevailing congestion. With each regime is associated a value function capturing driver preferences for gains associated with a particular acceleration. A probabilistic regime selection mechanism relates the driver's choices to prevailing traffic conditions. The model is calibrated against actual trajectory data. Initial results show that the model provides realistic behavioral patterns previously identified in the literature.