Multi-Agent Coordination across Diverse Applications: A Survey

TL;DR

This survey presents a unified framework for multi-agent coordination across diverse applications, answering what coordination is, why it is needed, who to coordinate with, and how to coordinate. It surveys three general MAS coordination tasks—coordinated learning, communication/cooperation, and conflict resolution—grounded in a framework that emphasizes dependency-based clustering and iterative system-level evaluation. Six MAS application domains are analyzed (SAR, warehouse/logistics, transportation, humanoid robots, satellites, and LLM-based MAS), illustrating how coordination principles are instantiated in varied settings. The paper identifies scalability, heterogeneity, and learning mechanisms, including LLM-based MAS, as key open challenges and highlights promising future directions such as hierarchical-decentralized hybrids and human-MAS collaboration to advance the field toward more capable, trustworthy distributed AI systems.

Abstract

Multi-agent coordination studies the underlying mechanism enabling the trending spread of diverse multi-agent systems (MAS) and has received increasing attention, driven by the expansion of emerging applications and rapid AI advances. This survey outlines the current state of coordination research across applications through a unified understanding that answers four fundamental coordination questions: (1) what is coordination; (2) why coordination; (3) who to coordinate with; and (4) how to coordinate. Our purpose is to explore existing ideas and expertise in coordination and their connections across diverse applications, while identifying and highlighting emerging and promising research directions. First, general coordination problems that are essential to varied applications are identified and analyzed. Second, a number of MAS applications are surveyed, ranging from widely studied domains, e.g., search and rescue, warehouse automation and logistics, and transportation systems, to emerging fields including humanoid and anthropomorphic robots, satellite systems, and large language models (LLMs). Finally, open challenges about the scalability, heterogeneity, and learning mechanisms of MAS are analyzed and discussed. In particular, we identify the hybridization of hierarchical and decentralized coordination, human-MAS coordination, and LLM-based MAS as promising future directions.

Figures (4)

• Figure 1: The number of publications and research areas of multi-agent system (MAS) research based on the records of Web of Science (WOS). The MAS topic covers 148 of a total of 252 research areas. The record count determines the rectangular size for each research area, with the top 15 being: Computer Science, Mathematics, Engineering, Automation Control Systems, Robotics, Telecommunications, Energy Fuels, Business Economics, Communication, Transportation, Instruments Instrumentation, Mathematical Computational Biology, Operations Research Management Science, Physics, and Oncology.
• Figure 2: The structure of this survey. A unified framework is introduced in Section \ref{['sec_framework']}. Coordination problems for general MAS are reviewed in Section \ref{['sec_mas_general']}. MAS applications are surveyed in Section \ref{['sec_application']}. Future and open research topics are discussed in Section \ref{['sec_challenge']}.
• Figure 3: The unified framework (perspective) of coordination in this survey. The coordination in sequential decision-making is an iterative process consisting of three components: evaluate the system-level performance, social choice on who to coordinate with, and how to coordinate. (Section \ref{['sec_framework']})
• Figure 4: General multi-agent tasks. (Section \ref{['sec_mas_general']})