Variable speed limit (VSL) control can enhance traffic efficiency, reduce accident risks, and improve speed harmonization. Traditional methods like rule-based and proactive approaches face limitations in adaptability, robustness, or computational complexity. Reinforcement Learning (RL) offers potential by enabling self-learning in complex environments to optimize traffic management. This paper proposes an improved Soft Actor-Critic (SAC) algorithm, incorporating multi-objective rewards, factorized action space, and weighted state mechanism to enhance VSL control. Using a Cellular Automation (CA) model to simulate driver behavior, effectiveness of VSL were evaluated in two cases: (1) varying driver compliance rates, and (2) truck-to-passenger ratios. Results show higher compliance significantly reduces bottleneck dissipation time, increases average speed, and lowers collision probability, with optimization improving as truck ratios decline.
Variable speed limit control method for highways based on an improved SAC algorithm
International Conference on Frontiers of Traffic and Transportation Engineering (FTTE 2024) ; 2024 ; Lanzhou, China
Proc. SPIE ; 13645 ; 136450P
2025-06-16
Conference paper
Electronic Resource
English
Analysis of traffic flow with Variable Speed Limit on highways
Springer Verlag | 2012
|Creating an intelligent variable speed limit system for rural highways
Automotive engineering | 1999
|Creating an Intelligent Variable Speed Limit System for Rural Highways
British Library Conference Proceedings | 1999
|Analysis of traffic flow with Variable Speed Limit on highways
Online Contents | 2012
|Applicability Analysis and Application Research of Variable Speed Limit Strategy on Highways
Springer Verlag | 2025
|