Distributed Algorithm for Selecting Leaders for Supervisory Robotic Swarm Control

Abstract

In this paper, we present a distributed algorithm for selecting multiple leaders in a swarm that can be used for supervisory control of the swarm system. The usage of optimally placed leaders within swarms can minimize the communication requirements for information dissemination across a robotic network. We formulate the leader selection problem as a combinatorial optimization problem and provide a novel characterization of the optimal solution based on the notion of Voronoi decomposition of a graph. Based on this characterization, we present an incremental, distributed algorithm to compute the leader set. We present simulation results to show that our approach results in optimal performance.