Consensus Of Multi-Agent Systems With Uncertain Communication

In recent years, the consensus problem of multi-agent systems has received much attention of many researchers. In the control field, the multi-agent consensus problem becomes one of the important problems in multi-agent coordination control. It is needed to design an appropriate protocol or control law such that the state of each agent converge to the same objective based on local information exchange between agents. The actual multi-agent systems are in various complex environment, therefore local information exchange between agents may be affected by uncertainties such as communication network switching and transmission delay, etc. This dissertation studies consensus problems of multi-agent systems with uncertain communication using algebra graph theory, matrix theory, control theory, stability theory and Markov jump system theory. The content of this dissertation is given in the following:The consensus problem is studied for mobile robot with Markov communication topology. Because it is difficult to know exactly the probability for every transition; the Markov chain considered has some unknown element in its transition probability matrix. Under a linear consensus protocol, a sufficient condition is presented for mean square consensus of mobile robots. A consensus protocol design method is provided based on linear matrix inequalities, bisection search and numerical optimization.The average consensus problem is studied in mean square sense for distributed inference in a wireless sensor network under Markovian communication topology of uncertain transition probability. A sufficient condition is presented for average consensus of linear distributed inference algorithm. Based on linear matrix inequalities and numerical optimization, a design method is provided of faster distributed inference systems.Under switching communication topology, the consensus problem of a class of multi-agent systems with nonlinear coupling is studied. By contraction analysis, consensus conditions are obtained of these systems.The consensus problem is studied for linear discrete time munti-agent systems with time-varying transmission delay. Markov communication topology and output feedback protocol are considered. Based on Lyapunov function, a protocol design method is given such that the multi-agent systems reach to mean square consensus.Mean consensus problem is studied for a class of discrete time multi-agent systems in which information exchange is subjected to some network-induced constraints. These constraints include package dropout, time delay and package disorder. Using Markov jump system method, the necessary and sufficient condition of mean square consensus is obtained and a design procedure is presented such that multi-agent systems reach mean square consensus.