Modeling Lane Preferences in Agent-Based Multi-Lane Highway Simulation

Simulation of highway driving was traditionally the domain of virtual physics based models. Yet, traffic simulation is incomplete without considering the drivers' conscious strategic and tactical behavior. These aspects can be naturally simulated through an agent-based driver model. In this paper, we describe a model of the strategic lane preferences of the drivers, with a special attention to the optimal lane positioning for a safe exit. Our experiments show that the simulated traffic of Orlando's Highway 408 matches well with the real world traffic data. The increased simulation detail can be applied to crash prediction and the control of intelligent transportation system devices, such as variable speed limits.