Game Theoretical Approach to Model Decision Making for Merging Maneuvers at Freeway On-Ramps

Drivers of merging vehicles decide when to merge by considering surrounding vehicles in adjacent lanes in their deliberation process. Conflicts between drivers of the subject vehicles (i.e., merging vehicles) in an auxiliary lane and lag vehicles in the adjacent lane are typical near freeway on-ramps. This paper models a decision-making process for merging maneuvers that uses a game theoretical approach. The proposed model is based on the noncooperative decision making of two players, that is, drivers of the subject and lag vehicles, without consideration of advanced communication technologies. In the decision-making process, the drivers of the subject vehicles elect to accept gaps, and drivers of lag vehicles either yield or block the action of the subject vehicle. Corresponding payoff functions for two players were formulated to describe their respective maneuvers. To estimate model parameters, a bi-level optimization approach was used. The next generation simulation data set was used for model calibration and validation. The data set defined the moment the game started and was modeled as a continuous sequence of games until a decision is made. The defined merging decision-making model was then validated with an independent data set. The validation results reveal that the proposed model provides considerable prediction accuracy with correct predictions 84% of the time.