Variable speed limit (VSL) control plays a crucial role in optimizing traffic efficiency. The connected and autonomous vehicles, as a controllable agent, can offer a potential method for multi-agent reinforcement learning-based variable speed limit strategy. To pursue the target between both traffic efficiency and safety, a multi-agent proximal policy optimization-based variable speed optimization strategy is proposed by leveraging connected and autonomous vehicles to form moving bottlenecks proactively (MAPPO-MVSL) in this paper. First, we construct a positive artificial moving bottleneck using connected and autonomous vehicles on a two-lane highway. Then, a Markov Process-based VSL strategy is presented, and MAPPO algorithm is applied to solve the problem. Compared with the baseline and the traditional single-agent variable speed control strategy, the proposed strategy improves traffic efficiency by 40%. Additionally, it enhances safety in the merge area, reducing danger by 16%.
Multi-agent Reinforcement Learning-Based Variable Speed Limit Strategy by Leveraging Connected and Autonomous Vehicles in Mixed Traffic Flow
Lect. Notes Electrical Eng.
International Conference on Autonomous Unmanned Systems ; 2024 ; Shenyang, China September 19, 2024 - September 21, 2024
Proceedings of 4th 2024 International Conference on Autonomous Unmanned Systems (4th ICAUS 2024) ; Chapter : 49 ; 547-557
2025-04-02
11 pages
Article/Chapter (Book)
Electronic Resource
English
Springer Verlag | 2022
|Taylor & Francis Verlag | 2023
|