Multi-Agent Distributed Coordination Control: Developments and Directions
College of Mechantronic Engineering and Automation, National University of Defense Technology（国防科技大学机电工程与自动化学院）；Research School of Engineering, the Australian National University（澳大利亚国立大学工程学院研究学院）
本文综述了分布式协调控制的发展，重点讨论了共识与编队控制，结合图论分析，并简要回顾了 rendezvous/alignment、swarming/flocking 和 containment control 等相关问题，最后探讨了实际应用中的研究方向。
摘要：
本文总结了分布式协调控制的最新发展，特别是共识和编队控制，图论在其中起核心作用，以呈现多智能体分布式协调控制的全面概述，并简要回顾了相关问题。在共识问题中，讨论了不同动态（一阶、二阶到高阶线性和非线性）智能体在不同通信条件下的共识结果，包括切换通信拓扑和时变延迟的情况，其中代数图论在协议设计、稳定性证明和收敛分析中非常有用。在编队控制问题中，回顾了代数图论在编队控制中的应用，重点讨论了刚性和持久图的发展。通过刚性和持久性概念，可以完成编队变换、分裂和重构，并设计出基于范围的编队控制律，以最少所需信息维持刚性/持久编队。随后简要介绍了 rendezvous/alignment、swarming/flocking 和 containment control 的最新结果，以呈现图论在协调控制问题中的综合观点。最后，针对实际应用，提出了一些值得进一步研究的协调控制方向。
英文摘要：
In this paper, the recent developments on distributed coordination control, especially the consensus and formation control, are summarized with the graph theory playing a central role, in order to present a cohesive overview of the multi-agent distributed coordination control, together with brief reviews of some closely related issues including rendezvous/alignment, swarming/flocking and containment control.In terms of the consensus problem, the recent results on consensus for the agents with different dynamics from first-order, second-order to high-order linear and nonlinear dynamics, under different communication conditions, such as cases with/without switching communication topology and varying time-delays, are reviewed, in which the algebraic graph theory is very useful in the protocol designs, stability proofs and converging analysis. In terms of the formation control problem, after reviewing the results of the algebraic graph theory employed in the formation control, we mainly pay attention to the developments of the rigid and persistent graphs. With the notions of rigidity and persistence, the formation transformation, splitting and reconstruction can be completed, and consequently the range-based formation control laws are designed with the least required information in order to maintain a formation rigid/persistent. Afterwards, the recent results on rendezvous/alignment, swarming/flocking and containment control, which are very closely related to consensus and formation control, are briefly introduced, in order to present an integrated view of the graph theory used in the coordination control problem. Finally, towards the practical applications, some directions possibly deserving investigation in coordination control are raised as well.
Silo-Bench: A Scalable Environment for Evaluating Distributed Coordination in Multi-Agent LLM Systems
Beijing University of Technology（北京理工大学）；Zhejiang University（浙江大学）；ETH Zürich（苏黎世联邦理工学院）；Meituan LongCat Interaction Team（美团LongCat交互团队）；Vector Institute for Artificial Intelligence（人工智能向量研究所）
The Internet of Agentic AI: Communication, Coordination, and Collective Intelligence at Scale
Department of Electrical and Computer Engineering（电气与计算机工程系）；NYU Center for Cybersecurity（纽约大学网络安全中心）；NYU Tandon School of Engineering（纽约大学坦顿工程学院）；New York University（纽约大学）
