Managing Context and Communication in Distributed Agentic UAV Swarms

2026-10-01

AI Summary

This study addresses the reasoning degradation caused by prolonged interaction histories and the prohibitive communication overhead of full-flooding protocols for small language models deployed in distributed UAV swarms. To this end, we propose an event-driven hierarchical memory and semantics-aware communication framework. Specifically, knowledge is structurally organized across core, local, and peer-level memories, while a deterministic gossip protocol based on semantic novelty is designed to enable selective information dissemination, replacing inefficient broadcasting. Simulations in search-and-rescue scenarios demonstrate that the proposed approach achieves a 100% task completion rate while reducing both inference token consumption and transmitted data volume by approximately 50% compared to baselines. Furthermore, it significantly decreases survivor counting errors, highlighting its effectiveness in enhancing collaborative intelligence under stringent resource constraints.

Abstract

Unmanned aerial vehicle (UAV) swarms increasingly rely on language-model agents to provide adaptive mission-level reasoning in uncertain environments. Fully distributed control, in which each UAV hosts an independent Small Language Model (SLM), removes reliance on a centralized coordinator but introduces an information-management problem: long-running interaction histories can degrade the reasoning context, while indiscriminate information dissemination increases communication and inference overhead. We address these challenges with a distributed UAV-agent architecture that enables continuous local SLM control through an event-driven reason-act-observe lifecycle. Runtime knowledge is represented as structured atomic notes and organized into core, local, and peer-specific memory. A deterministic interest-aware gossip engine selectively disseminates these notes according to recipient-specific semantic novelty and recency. We evaluate the architecture using ten UAVs in a simulated search-and-rescue mission. Our approach completes all experimental runs, whereas unrestricted flooding messages completes only 70-85\%, and delegating forwarding decisions to the SLM prevents mission completion in every run. Compared with unrestricted flooding, our approach approximately halves inference-token consumption, reduces transmitted data, and achieves lower survivor-count error.

Problem

• Distributed UAV swarms
• Small Language Models
• Context management
• Communication overhead
• Information dissemination

Innovation

• Distributed UAV Swarms
• Small Language Models
• Interest-aware Gossip
• Structured Memory
• Event-driven Architecture