Modeling Acceleration Decisions for Freeway Merges

Abstract

In uncongested traffic situations, a merge is executed when the available gap is sufficiently large. However, in congested traffic, acceptable gaps for merging are often not readily available, and merging can involve more complex mechanisms. For example, the driver in the target lane may slow down and cooperate with the merging driver, or the merging driver may become impatient and decide to force in, and compel the lag driver in the target lane to slow down. Choices of the merging plan or tactic affect the gap acceptance and acceleration decisions of the driver. A driver who has decided to force in, for instance, is likely to accept smaller gaps and accelerate to facilitate the merge. The chosen tactic at any instant, however, is not distinctly observable from the vehicle trajectory. The model presented in this paper extends previous research in modeling the effect of merging plans in the lane-changing decisions by integrating the acceleration decisions of the driver with the gap acceptance decisions. A combined model for merging plan choice, gap acceptance, target gap selection, and acceleration decisions of drivers merging from the on-ramp is developed in that regard. Parameters of all components of the models are estimated jointly with detailed vehicle trajectory data collected from Interstate 80 in California. The inclusion of the target gap choice and acceleration behavior components has been supported by a validation case study in which the model has been implemented in MITSIMLab and validated against the observed aggregate traffic data collected from US-101 in California.