There are many examples of multi-agent system in both nature and engineering, such as bird flocking, fish schooling, multi-robot systems, etc. Due to the broad potential applications, multi-agent systems have drawn more and more attention in the past decade. And the coordination control of multi-agent systems plays a key role in the research of multi-agent systems. In order to achieve the control objective, agents exchange information via a communication network, which is usually corrupted by communication noise, delays and other factors. The most common control objectives of coordination control of multi-agent systems include: consensus, containment, formation, coverage, etc. This thesis aims to study the consensus and containment control of multi-agent systems with communication noise. The contributions of this thesis are summarized as follows: The consensus problem of second-order integral multi-agent systems with communication noise and switching topologies is studied. And a sampled-data based consensus protocol is proposed for solving this problem. It is proved that if 1) all communication topologies are balanced and have spanning trees; 2) the time-varying gain satisfies the stochastic approximation type conditions; then the proposed protocol can solve the mean square consensus problem of second-order integral multi-agent systems with communication noise and switching topologies. Meanwhile, we also proved that the almost sure consensus can also be achieved by using the consensus protocol with some particular types of time-varying gains. The containment control problems of first-order and seconder-order integral multi-agent systems with communication noise is studied. Two types of containment protocols are designed for solving these two problems. It turns out that the proposed protocols can solve the mean square containment problems only if the following condition hold: for each follower there exists at least one directed path from one leader to this follower. Different from most papers, each agent is allowed to has its own time-varying gain. It is proved that if all time-varying gains are infinitesimals of the same order, the mean square containment can still be achieved. The consensus problem of multi-input multi-output multi-agent linear systems with communication noise and Markovian switching topologies is investigated. The state of each agent is assumed to be unavailable. To deal with this difficulty, the state observer is borrowed to estim...
修改评论