多智能体系统的分组一致性分析与实现

Researches on the coordination problems of multi-agent systems are booming in recent years. Interactions among different types of agents are receiving more and more attention, and the problem of especial interest is to let different groups of agents accomplish individual tasks via inter-group cooperation. This project is dedicated to the cluster consensus problem of multi-agent systems where agents belonging to the same cluster will reach consensus for their system states while not so for agents belonging to different clusters. Three main tasks will be accomplished in this project. The first task aims to analyze the role of interaction graphs in achieving cluster consensus, and to establish feasible graph topologies for cluster consensus. Both nonnegative interaction graphs and graphs with negative links will be investigated. The second task focuses on designing distributed control laws for linear multi-agent systems where the system models of the agents are nonidentical. The third task aims to design adaptive laws for the interaction strengths among agents, so as to avoid using information about the graph topology in the design of distributed control laws for the agents. The main results established in this project will be applied and demonstrated by experiments involving multiple ground vehicles and unmanned aerial vehicles. The results derived in this project can provide feasibility instructions and constructing methodologies for multi-agent systems consisting of multiple groups and multiple types of agents. Hence, research results of this project will contribute a lot to the completion of current multi-agent coordination theory.

随着多智能体系统协同理论的蓬勃发展，不同类型的智能体之间的交互受到越来越多的研究关注，尤其是让不同的群体既相互配合又分工完成各自的任务具有更实际的意义也更具有挑战性。本项目研究的多智能体系统分组一致性问题，目的是使属于同一分组的智能体实现其状态的一致，不同分组的智能体的状态一般不一致。本项目的研究任务有三个，其一是分析分组一致的可实现性对交互图的依赖关系，从而获得可行的交互图拓扑，研究对象包括非负交互图和含有负性连接的交互图；其二是针对异质的线性系统模型设计分布式的控制器；其三是研究交互图连通强度的自适应调节策略，以避免在控制器的设计中利用交互图的拓扑信息。相关主要理论结果将在多自主车辆、飞行器的协同编队任务中进行实验演示。本项目的研究成果将为多群体、多类型智能体系统的建立提供可行性指导及构建的方法支持，为其分组一致的实现提供理论与实验依据，从而进一步拓展并完善多智能体系统的协同理论。

围绕多智能体系统研究了其分组一致性的实现问题及其在编队控制和车队控制中应用时的基本问题。针对通过有向网络连接的线性多智能体系统，分析获得了网络拓扑结构和连通强度条件，设计了自适应分布式控制策略，为多智能体系统的构建和分组一致性控制的可实现性提供了判断依据和实现方法支持；针对异质的智能车辆编队，研究了其在通信信道衰减、传输延迟、网络攻击等情况下的控制策略和通信策略，保证了系统的鲁棒性；针对由无人机集群、自主机器人等组成的多智能体系统，提出了分层式协作方法和基于预测的运动规划算法，保证了多智能体系统在复杂环境下运动时的自主性和安全性。依托本项目共发表论文12篇，其中SCI检索期刊论文6篇、EI检索期刊论文2篇，另有4篇论文发表在IROS、ICCA、CCC等EI收录的国际国内会议。