Modeling the Dynamics of Lane-changing Durations

Inspired by detailed urban traffic data, we propose a generalized lane-changing model in which maneuvers are treated as discrete choice events with continuous execution times. In our approach maneuver duration emerges endogenously from kinematic, behavioral, and maneuverability constraints. The model explicitly distinguishes between gap acceptance and collision avoidance depending on the type of interaction with neighboring agents, generating realistic, collision-free trajectories. Monte Carlo simulations recover the empirically established lognormal distribution of lane-changing durations, providing the first mechanistic explanation for its emergence and significantly outperforming alternative distributions.