CARUS, an operational retasking application for a swarm of autonomous UAVs: First return on experience
Serge Chaumette, Rémi Laplace, Christophe Mazel, R. Mirault, Albert Dunand, Y. Lecoutre, J-N. Perbet · 2011
The goal of the CARUS1project is to demonstrate the effective operation of a swarm of UAVs. The objective is to keep under surveillance a given set of points of interest where incidents (such as a fire) can arise. Each individual UAV operates in an autonomous manner and the decisions are taken by each aircraft in the air rather than on the ground, based on asynchronous communication between vessels. These local decisions impact the whole swarm behavior. It should be noted that we are not addressing formation flights; we are dealing with groups of UAVs the goal of which is to cooperatively achieve a common mission. CARUS relies on the research activities carried out at the Laboratoire Bordelais de Recherche en Informatique (LaBRI) in the domain of secure swarms of autonomous mobile communicating terminals. It both takes advantage of the results of this research and contributes to some of its activities, among which a PhD funded by the DGA (Direction Générale de l'Armement). CARUS uses SCUAL2- pronounce [skwal]- which is a swarm of five UAVs (fig. 1) acquired thanks to a funding by the MIB3Carnot. It also relies on the UAV and ground station expertise of the Fly-n-Sense company and the support of Thales in terms of program methodology and of software for ground visualization of the swarm operation. CARUS is a LaBRI project, conducted in the framework of the Albatros GIS (Groupement d'Intérêt Scientifique), with the support of the Région Aquitaine, its UAV Cluster and Bordeaux TechnoWest. The goal of this paper is to give an overview of the CARUS demonstrator. Its development has raised several major issues that not only relate to distributed algorithms and embedded systems, but also to regulatory rules and safety procedures. The key point is that the target platform is a swarm of autonomous vessels and not a single UAV; it must thus be dealt with as a whole. For instance, when a member of the swarm has to be landed for security reasons, the other UAVs must be moved away from the landing path so as to avoid collisions. These are new problems that did not exist when one single vessel was classically considered.