多智能体系统分布式事件触发控制研究及其应用

Distributed cooperative control of multi-agent systems has recently received much attention from various scientific communities ranging from mechanics to applied mathematics, physics, and control science. Note that information interaction through a communication network among the agents is the basic way for them to cooperate. Given that the agents equip with limited energy sources and communication bandwidth, they can only exchange information at some discrete time instants. This project aims to study the characteristics and design of distributed event-triggered cooperative control schemes for avoiding continuous communication and reducing communication cost. Specifically, we want to establish fully distributed event-triggered cooperative controllers and to propose several robust event-triggered cooperative control laws in the presence of external disturbances, time delays, and uncertainties. Different from common design methods without consideration of robustness and dependent on a global coordinate frame or global information of the communication graph, the strategies proposed in this project will remove these limitations. Based on the above-mentioned works, distributed event-triggered control protocols will be further addressed for multiple mobile robots and manipulators. Especially, the corresponding event-triggered schemes will be developed by considering mechanical characteristics of these agents.

多智能体系统分布式协调控制是近十年来力学、应用数学、物理、控制等学科领域的交叉前沿课题。基于通讯网络的信息交互是多智能体系统相互协作的必要因素。但由于智能体自身能源或通信带宽的限制，智能体间只能在某些离散的时刻交互信息。为了避免连续通讯和降低通信损耗，本课题拟深入研究分布式事件触发协调控制策略的特征与设计，系统地提出并设计完全分布式的事件触发控制协议以及针对干扰、时滞和不确定性的鲁棒事件触发协调控制律，改进和突破现有的几类主流设计方法依赖全局坐标系或者网络拓扑全局信息、尚未考虑鲁棒性能等局限性。进一步结合多移动机器人以及多机械臂协调控制问题，给出相应的分布式事件触发控制协议设计，提出与控制对象特征相适应的事件触发策略。

本项目主要研究了多智能体系统的分布式事件触发控制。具体包括：针对已有的事件触发策略中存在触发条件上界过零问题，提出了相关的修正；设计了完全分布式的事件触发控制协议并在触发条件中引入了Lp函数，拓展了理论研究结果；提出了自适应分布式事件触发策略，解决了已有设计依赖于全局信息的问题；针对多智能体存在未知信息时，设计了分布式事件触发跟踪控制协议；开展了多智能体系统分布式事件触发刚性编队控制的研究；开展了一般线性/非系统的动态事件触发控制策略研究；开展了多无人机系统的分布式编队实验研究。本项目的研究完善了分布式事件触发控制的理论研究，为多智能体协同控制中如何有效减少信息交互和控制器更新提供了理论保障。