AI RESEARCH
A Robust and Efficient Multi-Agent Reinforcement Learning Framework for Traffic Signal Control
arXiv CS.AI
•
ArXi:2603.12096v1 Announce Type: new Reinforcement Learning (RL) in Traffic Signal Control (TSC) faces significant hurdles in real-world deployment due to limited generalization to dynamic traffic flow variations. Existing approaches often overfit static patterns and use action spaces incompatible with driver expectations. This paper proposes a robust Multi-Agent Reinforcement Learning (MARL) framework validated in the Vissim traffic simulator. The framework integrates three mechanisms: (1) Turning Ratio Randomization, a.