Education, Science, Technology, Innovation and Life
Open Access
Sign In
  1. Home
  2. Journals
  3. Automation and Machine Learning
  4. Vol 7, Issue 2, 2026

A Multi-Agent Reinforcement Learning-Based Collaborative Decision-Making Method for Intelligent Connected Vehicle Clusters in High-Density Traffic Scenarios

Download as PDF

DOI: 10.23977/autml.2026.070202 | Downloads: 2 | Views: 68

Author(s)

Quan Yu 1, Zhuo Bao 1

Affiliation(s)

1 School of Electrical and Control Engineering, North China University of Technology, Beijing, 100144, China

Corresponding Author

Zhuo Bao

ABSTRACT

To address the limitations of local observations, insufficient modeling of inter-vehicle interactions, and the difficulty of balancing traffic efficiency and operational stability in collaborative decision-making for intelligent connected vehicle clusters under high-density traffic conditions, this paper proposes a collaborative decision-making method based on multi-agent reinforcement learning. First, according to the operational characteristics of vehicle groups in high-density traffic scenarios, the collaborative decision-making requirements are clarified with the objectives of improving road resource utilization, average travel speed, and traffic flow stability. Second, the vehicle-cluster decision-making problem in high-density traffic scenarios is formulated as a decentralized partially observable Markov decision process. An explicit neighborhood interaction mechanism is introduced to characterize the dynamic local relationships among vehicles, and a structured reward function integrating traffic efficiency, operational stability, safety constraints, and rule compliance is designed. Furthermore, under the centralized training and decentralized execution framework, collaborative policy learning for vehicle clusters is achieved based on MAPPO. Finally, a high-density traffic simulation scenario is constructed on the highway-env platform, and comparative experiments are conducted against Rule-based, IPPO, and NI-MAPPO methods. The experimental results show that the proposed method effectively improves the collaborative traffic performance of vehicle clusters under high-density traffic conditions. In the test scenario, the average speed reaches 25.041 m/s, the speed standard deviation is 1.393 m/s, and the collision rate is 0, indicating favorable traffic efficiency, operational stability, and safety.

KEYWORDS

Intelligent connected vehicles; High-density traffic; Multi-agent reinforcement learning; Collaborative decision-making; MAPPO; Neighborhood interaction

CITE THIS PAPER

Quan Yu, Zhuo Bao. A Multi-Agent Reinforcement Learning-Based Collaborative Decision-Making Method for Intelligent Connected Vehicle Clusters in High-Density Traffic Scenarios. Automation and Machine Learning (2026). Vol. 7, No. 2, 9-15. DOI: http://dx.doi.org/10.23977/autml.2026.070202.

REFERENCES

[1] Tao S, Wen-Xing Z, Shi-Bin S, et al. Distributed sliding mode control strategy for intelligent connected vehicle platoon in complex media[J]. Applied Mathematical Modelling, 2024, 125(PB): 404-424. DOI: 10. 1016/J. APM. 2023. 10. 028.
[2] Fangwu M, Yu Y, Jiawei W, et al. Eco-driving-based cooperative adaptive cruise control of connected vehicles platoon at signalized intersections[J]. Transportation Research Part D, 2021, 92DOI: 10. 1016/J. TRD. 2021. 102746.
[3] Guichuan L, Rongchen Z, Ming S. Cooperative Adaptive Sliding Mode Platoon Control of Intelligent Connected Vehicles under Communication Interruption[J]. International Journal of Automotive Technology, 2023, 24(2): 513-525. DOI: 10. 1007/S12239-023-0043-4. 
[4] Wei Z, Dong C, Jun Y, et al. Multi-agent reinforcement learning for cooperative lane changing of connected and autonomous vehicles in mixed traffic[J]. Autonomous Intelligent Systems, 2022, 2(1): DOI: 10. 1007/S43684-022-00023-5. 
[5] Li S, “David” W F. Performance of State-Shared Multiagent Deep Reinforcement Learning Controlled Signal Corridor with Platooning-Based CAVs[J]. Journal of Transportation Engineering, Part A: Systems, 2023, 149(8): DOI: 10. 1061/JTEPBS. TEENG-7768. 
[6] Haotian S, Yang Z, Keshu W, et al. Connected automated vehicle cooperative control with a deep reinforcement learning approach in a mixed traffic environment[J]. Transportation Research Part C, 2021, 133DOI: 10. 1016/J. TRC. 2021. 103421. 
[7] Haotian S, Danjue C, Nan Z, et al. A deep reinforcement learning based distributed control strategy for connected automated vehicles in mixed trafficplatoon[J]. Transportation Research Part C, 2023, 148DOI: 10. 1016/J. TRC. 2023. 104019. 
[8] Pu Q, Xie K, Guo H, et al. Modeling interactive crash avoidance behaviors: A multi-agent state-space transformer-enhanced reinforcement learning framework. [J]. Accident; analysis and prevention, 2025, 226108334. DOI: 10. 1016/J. AAP. 2025. 108334. 
[9] Jinjue L, Chunhui Y, Zilin S, et al. A survey on urban traffic control under mixed traffic environment with connected automated vehicles[J]. Transportation Research Part C, 2023, 154DOI: 10. 1016/J. TRC. 2023. 104258. 
[10] Bellman R. Dynamic programming[M]. Princeton: Princeton University Press, 1957. 
[11] Sutton R S, Barto A G. Reinforcement learning: An introduction[M]. 2nd ed. Cambridge, MA: MIT Press, 2018. 
[12] Watkins C J C H, Dayan P. Q-learning[J]. Machine Learning, 1992, 8(3/4): 279-292. 
[13] Williams R J. Simple statistical gradient-following algorithms for connectionist reinforcement learning[J]. Machine Learning, 1992, 8(3/4): 229-256. 
[14] Konda V R, Tsitsiklis J N. Actor-critic algorithms[C]//Advances in Neural Information Processing Systems 13. Cambridge, MA: MIT Press, 2000. 
[15] Silver D, Lever G, Heess N, et al. Deterministic policy gradient algorithms[C]//Proceedings of the 31st International Conference on Machine Learning. Beijing: JMLR. org, 2014: 387-395. 
[16] Littman M L. Markov games as a framework for multi-agent reinforcement learning[C]//Proceedings of the 11th International Conference on Machine Learning. San Francisco: Morgan Kaufmann, 1994: 157-163. 

Subscription
E-Mail Alert
Most popular papers
Call for Special Issue
Publication Ethics & OA Statement
Plagiarism Policy
Contact Us
Downloads: 5361
Visits: 273443

Sponsors, Associates, and Links

All published work is licensed under a Creative Commons Attribution 4.0 International License.

Copyright © 2016 - 2031 Clausius Scientific Press Inc. All Rights Reserved.