Evaluating V2X Communications in the CARLA Simulator for a Collaborative Planning Use Case

Vehicle-to-Everything (V2X) communication is pivotal for enhancing the safety and efficiency of autonomous driving. The CARs Learning to Act (CARLA)simulation platform is a key tool for evaluating these technologies. This research implements a Robotic Operating System (ROS) module and an infrastructure server in CARLA to study V2X communication using Message Queuing Telemetry Transport (MQTT) protocol for a collaborative planning use case. An infrastructure module manages communication among the different agents of a traffic scene and a reactive route calculation system allows real-time route recalculation for each vehicle based on alerts received from the infrastructure and other vehicles. Evaluations of A* and Dijkstra planning algorithms reveal that A* is more efficient on small maps, while Dijkstra excels on larger maps. Our results validate the system's effectiveness, offering a foundation for future research in autonomous vehicles and traffic management.