Cooperative Perception of Connected Autonomous Vehicles: A Stackelberg Game Approach

The rise of connected autonomous vehicles demands a wider range of more accurate environment perception information. However, individual perception is hindered by blind spots and limited detection precision. Cooperative perception enhances vehicles’ environment perception capacity by sharing data with other vehicles or roadside infrastructure via vehicle-to-everything (V2X) communication, effectively addressing individual perception problems. Despite the critical importance of cooperative perception technology, the quality of transmitted data may decrease in environments with limited communication and computation resources, leading to a low coverage area and detection precision. To address the above issues and meet the perception utility demands of all vehicles in the environment, we propose a multi agent cooperative perception framework. Based on this framework, we propose a multi-leader multi-follower Stackelberg game model and design an iterative optimization algorithm to enhance perception utility while reducing system delay and overhead. Numerical results demonstrate that our proposed approach significantly improves the environment perception utility of vehicles compared to benchmark approaches.