Stability of leaderless discrete-time multi-agent systems (Q850839)

The paper extends the results concerning the relation between connectedness of the interaction graphs in multi-agent discrete-time systems and the ability for global convergence to a common equilibrium point. The multi-agent system is represented by a time-dependent graph describing the transfer of information between agents, and a set of rules describing updates of their states. The possibility of bounded time delays in the inter-agent communications and relaxing of the convexity assumption about admissible regions for the state transition mappings of the agents are pointed out. In particular, an original solution to rendezvous problems for populations of agents evolving in a nonconvex set is provided. Detailed formal analysis of the problem is presented. The specific feature of the approach is that the configuration of the information network is treated as an independent input, and the emphasis is put on the problem how the topology of interactions between agents affects the convergence of all agents to the same equilibrium. The analysis is carried out under the assumption of limited information transmission speed between agents. Illustrative examples are included.