Lane-Changing Decision Model of Connected and Autonomous Vehicles in Merging Area Based on Game Theory

Abstract

To enhance safety and comfort in merging zones characterized by complex interactive behaviors within mixed traffic flows of connected and autonomous vehicles (CAVs) and human-driven vehicles (HVs), this study develops a lane-changing decision model based on game theory for CAVs merging onto highways. A cooperation factor is introduced to quantify the impact of information exchange and facilitate smoother traffic flow interactions. Through the analysis of interactive behaviors, a lane-changing decision model is constructed, accommodating both CAV–CAV and CAV–HV interactions. The model is calibrated and validated through MATLAB simulations, with an emphasis on analyzing macrolevel traffic flow characteristics. Simulation results indicate that, when the cooperation factor is set to 0.4, traffic flow is optimized across varying mixing rates, highlighting the model’s effectiveness in managing mixed traffic scenarios. Additionally, the game lane-changing decision model shows a 12.5% increase in instances where the time-to-collision reciprocal is 0 at a 70% mixing rate, indicating potential areas for further safety improvement. Furthermore, at a traffic density of 20 veh/km with a 30% mixing rate, the throughput of the game decision model achieves 1,128 veh/h, representing a significant 17% improvement compared to the baseline scenario.