An Efficient Distributed Multivehicle Cooperative Tracking Framework via Multicast.

To support various upper applications of intelligent vehicles ranging from driving assistance to automated planning and control, accurate localization, and tracking are the fundamental tasks. Given the limited versatility and efficiency of traditional single-vehicle multisensor and multivehicle multisensor localization and tracking solutions, this article presents an efficient distributed multivehicle cooperative tracking framework via multicast. Once the self-positioning data is locally fused with assistance from roadside units, each vehicle shares the local-fusion results with surrounding vehicles through multicast and observes surrounding vehicles with on-board sensing equipment. The vehicles can then jointly feed the local-fusion results, received multicast information, and observation results into a global filter to obtain accurate and robust cooperative tracking. By leveraging multicast, the communication load is reduced, which promotes the efficiency of communication resource utilization. By optimizing the data fusion procedure, the error caused by error correlation is eliminated and the sensitivity to nonideal conditions, including packet loss, interruption, time-varying cooperative vehicles, etc., is reduced, which improves the versatility of the framework in real-world applications. Furthermore, several practical issues, such as random communication delay, packet loss, communication load, and localization robustness are also involved. To verify the effect of the framework, both theoretical analyses and simulation results are presented to show the accuracy and robustness of our proposed cooperative tracking framework.