---
格式版本: 2
标题: "SIGMA: Symmetry-aware, Intelligent, Geometric, Multi-objective Adaptive Control for Robust, Dependable Traffic Management"
原文链接: "https://arxiv.org/abs/2608.18263"
发布日期: "2026-08-18"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Tue, 18 Aug 2026 19:25:52 UTC (3,372 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-20T15:45:55+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-20T15:43:02+08:00"
入库时间: "2026-08-20T07:45:55.303Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=throughput&searchtype=all"
匹配关键词:
  - "throughput"
相关厂家:
  []
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 5
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "该资料为arXiv交通信号控制机器学习论文，仅命中throughput一词，与超节点/AI Rack/机柜级AI基础设施完全无关，属于明显无关内容。"
AI质检模型: "ali-deepseek-v4-flash"
AI质检时间: "2026-08-20T15:50:36+08:00"
AI主题相关性: 0
AI来源权威性: 5
AI新颖性: 0
AI技术细节: 0
AI商业部署信号: 0
AI完整性: 5
AI摘要: "SIGMA 提出一种结合大语言模型的多目标强化学习交通信号控制框架，将自然语言紧急指令转化为优先级向量，并用旋转增强提升路口间迁移能力。"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-09-07T03:22:22.953Z"
采集批次: "2026年8月20日14点19分32秒"
采集批次ID: "20260820-141932-079"
去重键: "https://arxiv.org/abs/2608.18263"
---

## Computer Science > Machine Learning

## Title:SIGMA: Symmetry-aware, Intelligent, Geometric, Multi-objective Adaptive Control for Robust, Dependable Traffic Management

Authors:[Pratham Payra](https://arxiv.org/search/cs?searchtype=author&query=Payra,+P), [Jagadish B](https://arxiv.org/search/cs?searchtype=author&query=B,+J), [Tanmay Sen](https://arxiv.org/search/cs?searchtype=author&query=Sen,+T), [Tanujit Chakraborty](https://arxiv.org/search/cs?searchtype=author&query=Chakraborty,+T)

[View PDF](https://arxiv.org/pdf/2608.18263) [HTML (experimental)](https://arxiv.org/html/2608.18263v1)

> Abstract:Traffic signal control is a complex sequential decision-making problem requiring real-time adaptation and trade-offs among throughput, delay fairness, signal stability, and emergency vehicle priority. Existing RL methods often fix objectives, ignore dynamic priority changes, and fail to generalize across geometrically similar [this http URL](http://intersections.we/) propose SIGMA (Symmetry-aware, Intelligent, Geometric, Multi-objective Adaptive traffic control), an RL framework enhanced with a large language model (LLM) for adaptive objective tuning and orientation-invariant learning. SIGMA converts natural-language emergency commands into priority vectors for a multi-objective actor-critic controller, avoiding manual reward engineering. Rotational augmentation improves transferability across four-way intersections, while offline-to-online learning ensures stable initialization and gradual adaptation to changing [this http URL](http://traffic.we/) define reliability properties covering emergency service levels, graceful degradation under LLM failures, and demand sensitivity, validated via bootstrap statistics. Evaluated in SUMO on four Kolkata-based urban intersections against fixed-time, actuated, and DQN controllers, SIGMA reduces average/emergency waiting times and queue lengths, and boosts throughput. Ablation studies confirm robustness to component failures and geometric rotations. Overall, SIGMA offers a reliable, language-guided, multi-objective traffic control system with statistical reliability assurance.

| Subjects: | Machine Learning (cs.LG); Machine Learning (stat.ML) |
| --- | --- |
| Cite as: | [arXiv:2608.18263](https://arxiv.org/abs/2608.18263) \[cs.LG\] |
|  | (or [arXiv:2608.18263v1](https://arxiv.org/abs/2608.18263v1) \[cs.LG\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2608.18263](https://doi.org/10.48550/arXiv.2608.18263) |

## Submission history

From: Pratham Payra \[[view email](https://arxiv.org/show-email/1e0a6871/2608.18263)\]  
**\[v1\]** Tue, 18 Aug 2026 19:25:52 UTC (3,372 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2608.18263) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))