Argus: Realistic target coverage by drones

Abstract

Low-cost mini-drones with advanced sensing and maneuverability enable a new class of intelligent visual sensing systems. This potential motivated several research eforts to employ drones as standalone surveillance systems or to assist legacy deployments. However, several fundamental challenges remain unsolved including: 1) Adequate coverage of sizable targets; 2) Target orientation that render coverage efective only from certain directions; 3) Occlusion by elements in the environment, including other targets. In this paper, we present Argus, a system that provides visual coverage of wide and oriented targets, using camera-mounted drones, taking into account the challenges stated above. Argus relies on a geometric model that captures both target shapes and coverage constraints. With drones being the scarcest resource in Argus, we study the problem of minimizing the number of drones required to cover a set of such targets and derive a best-possible approximation algorithm. Building upon that, we present a sampling heuristic that performs favorably, while running up to 100x faster compared to the approximation algorithm. We implement a complete prototype of Argus to demonstrate and evaluate the proposed coverage algorithms within a fully autonomous surveillance system. Finally, we evaluate the proposed algorithms via simulations to compare their performance at scale under various conditions.