Coverage control by multi-robot networks with limited-range anisotropic sensory

Abstract

This article considers the deployment of a network of robotic agents with limited-range communication and anisotropic sensing capabilities. We encode the environment coverage provided by the network by means of an expected-value objective function. This function has a gradient which is not amenable to distributed computation. We provide a constant-factor approximation of this measure via an alternative aggregate objective function, whose gradient is spatially distributed over the limited-range Delaunay proximity graph. We characterise the smoothness properties of the aggregate expected-value function and propose a distributed deployment algorithm that enables the network to optimise it. Simulations illustrate the results.