Decentralised coordination of a multi-agent system based on intermittent data

Abstract

In this paper, we present a novel decentralised and non-cooperative algorithm for estimation and control of a multi-agent system. The control goal is to achieve a balanced formation on a generic closed curve. Different from previous work, each agent only gathers a measurement of its Euclidean distance from the other agents when they are in its proximity. This distance is usually different from the controlled distance along the curve, thus producing an uncertain and intermittent information on the actual spacing among agents. This fleeting data flow is processed by an estimation algorithm to produce an interval estimate of the relative position, which is then used by an ‘interval feedback control law’ to steer the system dynamics. The effectiveness of the approach and its performance are demonstrated through an extensive numerical analysis on two representative examples.