Research On Swarming Behavior Of Multi-agent Systems In Networks

Recently, the research on swarming behaviour of multi-agent system in network is an active area which make strong attention of scientists and engineers in various fields,such as computer science, intelligent control, network communication, mathematics, social sciences and so on. This is mainly due to broad applications of multi-agent system in many areas including cooperative control of unmanned air vehicles(UAVs), formation control of multi-robot, distributed sensor networks, attitude alignment of clusters of satellites, and congestion control in communication networks. This paper studies swarming behaviour of multi-agent system in network which has some nonlinear properties using graph theory, matrix analysis, control systems and other theories and methods. The main results of this paper are stated as follows:1.The stability analysis of multi-agent systems in network with some nonlinear properties and with and/or without time delay in communication network is studied in this paper. It shows that the agents of multi-agents system in indirect communication network will asymptotically form a cohesive cluster with finite size if the nonlinear attraction and repulsion functions satisfy some very mild assumptions. In comparison with the existing literature, we generalise the attraction / repulsion function into a class of nonlinear functions with some properties in multi-agent network system, which reveals some more rich self-organization dynamics of the network system.2. The coordination control of multi-agent systems with some nonlinear properties under the influence of environment, and with or without time delay are discussed in this paper. For reciprocal multi-agent systems with some nonlinear properties under the influence of environment and with or without time delay, all the agents will aggregate and eventually form a cohesive cluster of finite size for different profiles under some very mild assumptions. Furthermore, for the quadratic attractant/repellent profiles and the Gaussian attractant/repellent profiles, it shows that all the agents will move toward some favorable areas of the profile and diverge from unfavorable areas along the statecenter of system. For general non-reciprocal networks, our simulation shows that the members of the system with some nonlinear properties could appear some oscillatory behavior depending on the choices of adjacency matrix and different profiles.3. The coordination control of multi-agent systems with leaders and followers under the influence of environment in the two case that with or without time delay is presented in this paper. For reciprocal multi-agent systems with leaders and followers under the influence of environment in the case of no time delay, it is shown that under certain assumptions the follower agents can accurately reach the target or away from the target in the absence of environmental information by the navigation of leader agents.The size of the bounded area when the multi-agent system aggregates under different profiles is estimated. The size of the bounded area depends on the value of the parameters in the environment and the attraction repulsion function etc.. For reciprocal multi-agent systems with leaders and followers under the influence of environment in the case of with time delay, similar conclusions are also obtained. Numerical simulations are given to demonstrate the validation of the analytical results.