Consensus Within, Competition Between: Decentralized LLM Teams for Perimeter Defense
Consensus Within, Competition Between: Decentralized LLM Teams for Perimeter Defense
This research paper investigates the architectural trade-offs between centralized and decentralized large language model (LLM) teams in a simulated perimeter-defense scenario against LLM-controlled intruders. The authors compare three approaches: a central commander, an auction-based consensus team, and independent agents across five network conditions. Results indicate that the consensus team achieved the lowest breach rate of 15%, compared to 32% for the commander and 29% for independent agents. The study further explores how limited radio range affects performance and how visual input helps distinguish civilian drones from intruders, noting that a one-sentence forewarning in the prompt significantly reduces misidentification rates.
The study compares centralized, consensus-based, and independent LLM agent architectures in a perimeter-defense simulation, reporting specific breach rates under varying network conditions and noting that adding visual data for civilian drones improved discrimination only when explicitly prompted.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Environment
- SimulationOpen source S2
- code: Available Artifact link Open source S5
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
arXiv:2610.41044v1 Announce Type: cross 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 intrude
Open source S2
rs, 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 computa
Open source S3
tion 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
Open source S4
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: https://github.com/the-kas-lab/consensus-within-competition-between.
Open source S5
Source：arXiv Robotics — research abstracts · arxiv.org