On rendezvous control with randomly switching communication graphs

In this paper we analyze randomized coordination control strategies for the rendezvous problem of multiple agents with unknown initial positions. The performance of these control strategies is measured in terms of three metrics: average relative agents’ distance, total input energy consumption, and number of packets per unit time that each agent can receive from the other agents. By considering an LQ-like performance index, we show that a-priori knowledge about the first and second order statistics of agents’ initial position can greatly improve performance as compared to rendezvous control strategies based only on relative distance feedback. Moreover, we show that randomly switching communication topologies, as compared to static communication topologies, require very little information exchange to achieve high performance even when the number of agents grows very large.