Please use this identifier to cite or link to this item: https://hdl.handle.net/11264/62
Title: Implementing Autonomous Unmanned Aerial Vehicle Tactics on Quadrotor Aircraft Using Linear Model Predictive Control
Authors: Iskandarani, Mohamad
Royal Military College of Canada / Collège militaire royal du Canada
Givigi, Sidney
Keywords: Autonomous robotics
Cooperative robotics
Unmanned aerial vehicle
Linear model predictive control
Issue Date: 8-May-2014
Abstract: UAVs are gaining great interest due to their wide area of applications in the military and civilian fields. Applying these UAVs autonomously and in a decentralized manner to the accomplishment of various tasks is a growing trend in the field of cooperative robotics. Among these challenging group tasks we find certain UAV tactics, such as dynamic encirclement and formation flights, operating under the umbrella of safe and robust tactic switching. Dynamic encirclement is defined as the situation in which a target is isolated and surrounded by a UAV team in order to maintain awareness and containment of it. Formation flights allow for line abreast, triangle or cross formations while in flight, during which the team members match distance and speed with the others. In this thesis, the problem of switching from formation flights to dynamic encirclement is considered, and a decentralized Linear Model Predictive Control (LMPC) strategy is formulated. Other control techniques such Taylor series linearization (TSL) and Feedback Linearization (FL) are used to linearize the complex tactics models. It is shown, through simulation results and experimental validation, that the designed control policy is effective for a team of N UAVs in formation flights switching to dynamic encirclement of a stationary and moving target. With real-world flight tests, the Qball-X4 quadrotor aircraft is used as a validation platform.
Les drones, appliqu és aux domaines civiles et militaires, ont suscit és un int erêt croissant à cause de leur potentiel. L'utilisation des drones, d'une fa çon autonome et d écentralis ée, pour l'accomplissement de taches vari ées est une tendance croissante dans la communaut é de la robotique coop erative. Parmi ces taches complexes, on trouve des tactiques de drones, tels que l'encerclement dynamique et les vols en formations, o u les v éhicules accomplissent un transfert stable et robuste d'une tactique à l'autre. Limiter le mouvement et avoir conscience d'une cible est consid er é comme partie int egrante de l'encerclement dynamique. Les vols en formations, tels que ligne de front, triangle et croix, permettent aux membres de l' équipe de maintenir les distances et les vitesses requises par rapport aux autres. Dans ce m emoire, le probl ème de transfert stable des vols en formation à l'encerclement dynamique est consid er é et une strat egie de commande pr édictive lin eaire est formul ée. Autres techniques de controles, telles que la lin earisation par s eries de Taylor et la lin earisation par r etroaction, sont employ ées pour simplifi er les mod èles non-lin eaires repr esentants les tactiques. Il est d emontr é, à travers des r esultats en simulation et des validations exp érimentales, que la politique de controle propos ée est e fficace pour une equipe de N drones qui changent de vols en formation à encerclement dynamique. Par rapport aux vols exp erimentaux, le quadrirotor Qball-X4 est utilis é.
URI: https://hdl.handle.net/11264/62
Appears in Collections:Theses

Files in This Item:
File Description SizeFormat 
Thesis_Iskandarani.pdf3.47 MBAdobe PDFThumbnail
View/Open


Items in eSpace are protected by copyright, with all rights reserved, unless otherwise indicated.