Coverage Control of a Mobile Multi-Agent Serving System in Dynamical Environment

Abstract

This paper provides a solution for coverage control of a mobile multi-agent system, which needs to serve in a dynamical environment containing different importance in its sub-areas with varying resources, by proposing a model predictive broadcast control scheme. The broadcast control (BC) is a novel method for stochastic optimization of a multi-agent system for a global task, e.g., coverage control, without complex and costly agent-to-agent communication but through undistinguished broadcast signal from a coordinator agent. For large multi-agent systems, the existing BC methods are not highly efficient and often fall into local optimal. To overcome such a drawback, a model predictive BC scheme is proposed, where the individual agent makes predictive moves in a multi-step horizon and determines a better deterministic action based on multiple signals from the coordinator. The proposed model predictive BC is applied to the modified coverage problem of the mobile multi-agent systems, and the performance is compared with the existing BC methods in three different case studies. It is found that the proposed scheme outperforms the other BC schemes despite a large number of agents in the system.