arXiv cs.AI· Khawaja Abaid Ullah, Ananya Acharya, Mohammad Javad Khojasteh·· 1 天前AI 评分34
去中心化 LLM 团队如何用于周界防御：共识机制与集中指挥的对比
Consensus Within, Competition Between: Decentralized LLM Teams for Perimeter Defense
AI 导读
在简化的周界防御模拟中，采用拍卖式共识机制的 LLM 团队突破率为 15%，低于集中指挥官的 32% 和独立智能体的 29%。五种网络条件下仅有限无线电范围会改变团队突破率，且让低成本 LLM 更多参与投标计算会使突破率上升 38 个百分点。面对伪装成标本鸟无人机的入侵者，近前沿模型在 59-61% 的决策中误判为鸟类而不追击，提示词中加入一句预警可将其降至 0-25%。
正文 · 原文
Abstract:The use of large language models (LLMs) to coordinate multi-agent teams raises a fundamental architectural question: whether task allocation should be centralized in a single LLM controller or distributed across autonomous LLM agents, particularly when communication is unreliable. In a deliberately simplified perimeter-defense simulation against LLM-controlled intruders, we compared a central commander, an auction-based consensus team and independent agents under five network conditions. Breach rates were 15% for the consensus team, 32% for the commander and 29% for independent agents, and of the different network conditions, only limited radio range changed a team's breach rate, raising the commander's. In consensus team, giving a low-cost LLM more authority over the bid computation raised breaches by 38 percentage points. With birds and civilian drones in airspace, we added a photograph for each to the input text state. Only the near-frontier model treated them apart from intruders. Intruders disguised as existing taxidermy bird drones were mistook as birds by both low-cost and near-frontier LLMs, going unpursued in 59-61% of near-frontier decisions (1-16% before). However, a one-sentence forewarning in the prompt lowered it to 0-25%. We find that LLM team with a dedicated external conflict resolution protocol results in fewest breaches. Code: this https URL.
| Comments: | 88 pages (33 main text and references, 55 supplementary material), 7 main figures. Code: this https URL |
| Subjects: | Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Robotics (cs.RO) |
| Cite as: | arXiv:2610.41044 [cs.AI] |
| (or arXiv:2610.41044v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.41044 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Khawaja Abaid Ullah [view email]
[v1]
Thu, 29 Oct 2026 22:34:13 UTC (2,771 KB)
来源：arXiv cs.AI · arxiv.org