A Framework for Programming a Swarm of UAVs

Abstract

In recent years, sensing systems in urban environments are being replaced by Unmanned Aerial Vehicles (UAVs). UAVs, also known as drones, have shown great potential in executing different kinds of sensing missions, such as search and rescue, object tracking, inspection, etc. The UAVs' sensing capabilities and their agile mobility can replace existing complex solutions for such missions. However, coordinating a swarm of drones for mission accomplishment is not a trivial task. Existing works in the literature focus solely on managing the swarm and do not provide options for automating entire missions. In this paper, we present PaROS (PROgramming Swarm), a novel framework for programming a swarm of UAVs. PaROS provides a set of programming primitives for orchestrating a swarm of drones along with automating certain types of missions. These primitives, referred as abstract swarms, control every drone in the swarm, hiding the complexity of low level details from a programmer such as assigning flight plans, task partitioning, failure recovery and area division. Our experimental evaluation proves that our approach is stable, time-efficient and practical.