Research On Distributed Formation Algorithm For Networked Multi-agent Systems

With the spatial distribution, task parallelism, system scalability and strong faulttolerance ability, Multi-agent systems (MASs) has been received the high attention fromthe academic circles to the industry communities, and have been widely used in military,manufacture, livelihood and many other fields. Formation control has been thefundamental problems in MAS research as well as the theory premise of manyapplications of MAS. To meet more application requirements we further focus on theproblems of communication delay, message packet dropout, and topology change andenergy dissipation. Main content as follows:Firstly, this thesis through design distributed velocity observer investigates theformation problem for second-order continuous-time multi-agent systems under the casesthat witch switching topology and part of the leader information can not be measured, thenillustrated that the agents with the designed controllers and observers can achieve andmaintain the desired formation within a dynamic environment. By using the Lyapunov-Razumikhin theory, the upper bound of the varying time delays is given to ensure MASstable and realize desire formation.Secondly, the problems of autonomic formation control and topology optimizationfor second-order continuous-time multi-agent systems with time varying topologies andwaster of energy are investigated. Supposing that the initial network topology is connected,we design formation control strategy based on potential function, analyze the system’sstability by Lasalle’s invariant set theory. And then, a distributed topology optimizationscheme is designed based on optimally rigid graph, which successfully cuts down thecommunication links and reduces the energy consumption.Finally, owing to the inconsistence between researched model and application model,we transform second-order continue system to second-order discrete system, and then theleader-follower formation problem can be researched in this system. We give theconditions to finish formation task under the fixed and switched topology situation, andgive the sufficient and necessary condition that sampling interval must meet.