arXiv cs.MA· Lu Liu, Chi Xie, Xi Xiong·· 6 天前AI 评分22
基于物理信息世界模型的离线多智能体强化学习用于协同混合交通控制
Offline Multi-Agent Reinforcement Learning with a Physics-Informed World Model for Cooperative Mixed Traffic Control
AI 导读
研究提出一种物理信息世界模型驱动的离线多智能体强化学习框架，从局部 CAV 观测-动作历史重建可解释的全局交通状态，并以宏观-微观耦合交通动力学提供物理监督。概率集成世界模型学习状态转移与系统奖励，模型分歧量化认知不确定性，多步想象推演结合悲观奖励与不确定性截断用于离线策略学习。在 SUMO 匝道瓶颈场景约 1×10^6 条离线转换数据上，物理监督提升了状态重建与世界模型预测精度。
正文 · 原文
Abstract:This study investigates cooperative control of connected and automated vehicles (CAVs) at partially observable highway bottlenecks in mixed traffic, aiming to mitigate congestion without relying on complete global traffic states or online trial-and-error. We propose a physics-informed world model-based offline multi-agent reinforcement learning framework that reconstructs a physically interpretable global traffic state from local CAV observation-action histories, with coupled macroscopic-microscopic traffic dynamics providing physics-based supervision. A probabilistic ensemble world model learns traffic-state transitions and system rewards, while model disagreement quantifies epistemic uncertainty. Multi-step imagined rollouts with pessimistic rewards and uncertainty-driven truncation are then used for offline policy learning. Experiments in a SUMO-based on-ramp bottleneck using approximately $1\times10^6$ offline transitions show that physics supervision improves state reconstruction and world-model prediction accuracy.
| Subjects: | Multiagent Systems (cs.MA) |
| Cite as: | arXiv:2608.17739 [cs.MA] |
| (or arXiv:2608.17739v2 [cs.MA] for this version) | |
| https://doi.org/10.48550/arXiv.2608.17739 arXiv-issued DOI via DataCite |
Submission history
From: Lu Liu [view email]
[v1]
Tue, 18 Aug 2026 13:03:51 UTC (2,797 KB)
[v2]
Fri, 9 Oct 2026 02:59:13 UTC (2,789 KB)
来源：arXiv cs.MA · arxiv.org