Persistent Airborne Surveillance using Semi-Autonomous Drone swarms

Abstract

Persistent surveillance is an essential tool in many damage prevention and mitigation scenarios. Existing approaches to conduct surveillance over an area either require deploying static equipment on the ground or are expensive and have a high ecological footprint. The era of commercial drones has facilitated the development of lightweight, agile vehicles that can carry a wide variety of payloads with a very low ecological footprint. In this paper, we present a new airborne surveillance system using drone swarms. We explore the concept of Swarm Utility and present a mobility model for persistent surveillance that achieves maximum utility while being energy efficient. We program the drones in the swarm to follow trajectories defined by the model, eliminating the need to control the drones manually. Additionally, we augment our system with an object detection framework that runs on accumulated video feed streamed by the swarm to detect objects of interest in the area with high accuracy.