Distributed Model Predictive Control of Connected and Automated Vehicles With Markov Packet Loss

• Published in: IEEE transactions on transportation electrification Vol. 11; no. 2; pp. 6368 - 6379
• Main Authors: Bian, Yougang, Wang, Xuan, Tan, Yan, Hu, Manjiang, Du, Changkun, Sun, Zhongqi, Guo, Ge
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.04.2025; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Asymptotic stability; Automation; Communication; Connected and automated vehicle (CAV); consensus; Constraints; Controllers; Convergence; Cooperative control; distributed model predictive control (DMPC); Markov packet loss; Markov processes; Numerical stability; Packet loss; Platooning; Predictive control; Stability criteria; State feedback; Switches; Symmetric matrices; Systems stability; Topology; Trajectory; Transportation; vehicle platoon
• ISSN: 2332-7782; 2577-4212; 2332-7782
• DOI: 10.1109/TTE.2024.3507864

Communication packet loss is omnipresent and can disrupt state convergence and weaken system stability for connected and automated vehicles (CAVs). Research into communication packet loss has primarily focused ON state feedback, often leaving various constraints and objectives unaddressed. This study introduces a novel distributed model predictive control (DMPC) method for cooperative control of a CAV platoon subject to Markov communication packet loss. First, the vehicle platoon system is modeled, within which the state of each communication link is described by a Markov process. Second, a DMPC controller is designed by formulating an online open-loop optimization problem. Specifically, a self-deviation constraint is adopted to enhance robustness against packet loss, and a terminal state constraint with a consensus protocol-based update law is designed to achieve terminal mean-square consensus. Third, the terminal consensus, recursive feasibility, and internal stability are analyzed, and a sufficient condition on the asymptotic mean-square stability is deduced. A modified string stable DMPC is further designed with leader information. Finally, numerical simulations and an experiment are carried out to validate the proposed methods, revealing that the proposed DMPC controller outperforms the benchmark controller on tracking performance and fuel economy.