An Agent-Based Model for Evaluating Connected and Autonomous Vehicles, Collective Behaviors and Traffic System Performance

The rapid advancement of Connected and Autonomous Vehicles (CAVs) has the potential to revolutionize transportation systems by improving traffic efficiency, safety, and adaptability. While many studies have examined the performance, behavior, and decision-making of individual vehicles or small groups, evaluating the performance of large-scale traffic systems with CAVs remains underexplored. This study presents an agent-based modeling (ABM) framework to examine the CAVs dynamics and evaluate overall traffic system performance by simulating individual vehicle behavior and their dynamic interactions. The model focuses on decentralized vehicle decision-making, where each CAV independently perceives its surroundings, exchanges information via Vehicle to Vehicle (V2V) communication, and dynamically adjusts its speed and route based on real-time traffic conditions. The simulation is conducted on the road network of Katy, Texas, extracted from OpenStreetMap, with vehicles navigating dynamically through entry, exit, and internal road segments. Results demonstrate the proposed model has the capability of simulating various CAVs traffic systems at different scenarios. Furthermore, simulation findings suggest that an appropriate approach of decentralized decision-making enhances traffic stability and adaptability, underscoring this model's value for assessing CAV coordination strategies in urban environments.