[论文解读] Large Language Models for Multi-Robot Systems: A Survey

Peihan Li, Zijian An+2 more|ArXiv.org|Feb 6, 2025

Topic Modeling被引用 4

一句话总结

本文综述大语言模型（LLMs）如何在多机器人系统（MRS）中整合，覆盖高层、中层、低层规划以及人机交互，重点讨论架构、应用、挑战与未来方向。

ABSTRACT

The rapid advancement of Large Language Models (LLMs) has opened new possibilities in Multi-Robot Systems (MRS), enabling enhanced communication, task planning, and human-robot interaction. Unlike traditional single-robot and multi-agent systems, MRS poses unique challenges, including coordination, scalability, and real-world adaptability. This survey provides the first comprehensive exploration of LLM integration into MRS. It systematically categorizes their applications across high-level task allocation, mid-level motion planning, low-level action generation, and human intervention. We highlight key applications in diverse domains, such as household robotics, construction, formation control, target tracking, and robot games, showcasing the versatility and transformative potential of LLMs in MRS. Furthermore, we examine the challenges that limit adapting LLMs in MRS, including mathematical reasoning limitations, hallucination, latency issues, and the need for robust benchmarking systems. Finally, we outline opportunities for future research, emphasizing advancements in fine-tuning, reasoning techniques, and task-specific models. This survey aims to guide researchers in the intelligence and real-world deployment of MRS powered by LLMs. Based on the fast-evolving nature of research in the field, we keep updating the papers in the open-source GitHub repository.

研究动机与目标

• Motivate the use of LLMs to address coordination, scalability, and real-world adaptability in MRS.
• Categorize LLM applications in MRS into high-level task planning, mid-level motion planning, low-level action generation, and human-robot interaction.
• Identify communication architectures and decision-making paradigms for LLM-enabled MRS.
• Highlight current challenges (reasoning limits, hallucination, latency) and propose future research directions.

提出的方法

• Survey and synthesis of recent works integrating LLMs into MRS from the literature.
• Categorization of applications by planning level (high, mid, low) and human intervention.
• Analysis of communication architectures for LLMs in embodied multi-robot settings (centralized, decentralized, hybrid).
• Discussion of techniques to enhance LLM performance in MRS, such as fine-tuning (LoRA), retrieval-augmented generation (RAG), and multimodal capabilities.
• Comparison of frameworks and prompts used to decompose tasks and allocate to heterogeneous robot teams.

Figure 1: Overview of the applications of LLMs in MRS as introduced in Sec. 4 .

实验结果

研究问题

• RQ1 What are the main application categories where LLMs enhance MRS performance?
• RQ2 What communication architectures and prompting strategies best support coordination among multiple robots?
• RQ3 What are the primary challenges preventing reliable deployment of LLMs in MRS and what are potential solutions?
• RQ4 How do high-/mid-/low-level planning tasks interrelate when powered by LLMs in MRS?
• RQ5 What benchmarks and environments exist for evaluating LLM-enabled MRS systems?

主要发现

• LLMs enable improved communication, task planning, and human-robot collaboration in MRS.
• Four application categories are identified: high-level task allocation, mid-level motion planning, low-level action generation, and human intervention.
• Hybrid and centralized/decentralized architectures (e.g., HMAS-2, CMAS, DMAS) show different scalability and efficiency trade-offs across tasks.
• Techniques like LoRA, RAG, and multimodal LLMs are highlighted as important for domain adaptation and real-time decision making.
• There is ongoing need for benchmarks and simulation environments to evaluate LLM-enabled MRS systems.

Figure 2: The BOLAA architecture, which employs a controller to orchestrate multiple LAAs [ 59 ] .