The use of autonomous UAV’s is a progressively expanding industry. This thesisfocuses on the landing procedure with the main goal to be independent of visualaids. That means that the landing site can be hidden from the air, the landingcan be done in bad weather conditions and in the dark. In this thesis the use ofradio signals is investigated as an alternative to the visual sensor based systems.A localization system is needed to perform the landing without knowing wherethe landing site is. In this thesis an Extended Kalman Filter (EKF) is derived andused for the localization, based on the received signal strength from a radio beaconat the landing site. There are two main goals that are included in the landing,to land as accurate and as fast as possible. To combine these two goals a simultaneoustrajectory optimization and target estimation problem is set up that can bepartially solved while flying. The optimal solution to this problem produces thepath that the UAV will travel to get the best target localization while still reachingthe target. It is shown that trying to move directly towards the estimated landingsite is not the best strategy. Instead, the optimal trajectory is a spiral that jointlyoptimizes the information from the sensors and minimizes the arrival time.

This thesis concerns with the design and development of automatic flight controller strategies for the autonomous landing of fixed wing unmanned aircraft subject to severe environmental conditions. The Tactical Unmanned Aerial Vehicle (TUAV) designed at the Middle East Technical University (METU) is used as the subject platform. In the first part of this thesis, a dynamic model of the TUAV is developed in FORTRAN environment. The dynamic model is used to establish the stability characteristics of the TUAV. The simulation model also incorporates ground reaction and atmospheric models. Based on this model, the landing trajectory that provides shortest landing distance and smallest approach time is determined. Then, an automatic flight control system is designed for the autonomous landing of the TUAV. The controller uses a model inversion approach based on the dynamic model characteristics. Feed forward and mixing terms are added to increase performance of the autopilot. Landing strategies are developed under adverse atmospheric conditions and performance of three different classical controllers are compared. Finally, simulation results are presented to demonstrate the effectiveness of the design. Simulation cases include landing under crosswind, head wind, tail wind, wind shear and turbulence.

This thesis presents the design of an autopilot and guidance system for an unmanned aerial vehicle. Classical (PID) and modern control (LQT, Sliding Mode) methods for autonomous navigation and landing in adverse weather conditions are implemented. Two different guidance systems are designed in order to navigate through waypoints during normal and/or emergency flight. The nonlinear Pioneer UAV model is used in controller development and simulations. Aircraft is linearized at different trim points and total airspeed, altitude, roll and yaw autopilots are designed using Matlab/Simulink environment for lateral and longitudinal control of the aircraft. Gain scheduling is used to combine controllers designed for different trim points. An optimal landing trajectory is determined using &ldquo / Steepest Descent&rdquo / Algorithm according to the dynamic characteristics of the aircraft. Optimal altitude trajectory is used together with a lateral guidance against cross-wind disturbance. Finally, simulations including landing under crosswind, tailwind, etc., are run and the results are analyzed in order to demonstrate the performance and effectiveness of the controllers.

Thesis (MScEng)--Stellenbosch University, 2013. / ENGLISH ABSTRACT: This dissertation presents the design and practical demonstration of a flight control system (FCS) that is capable of autonomously landing a fixed-wing, unmanned aerial vehicle (UAV) on a stationary platform aided by a high-precision differential global positioning system. This project forms part of on-going research with the end goal of landing a fixed-wing UAV on a moving platform (for example a ship’s deck) in windy conditions. The main aim of this project is to be able to land the UAV autonomously, safely and accurately on the runway. To this end, an airframe was selected and equipped with an avionics payload. The equipped airframe’s stability derivatives were analysed via AVL and the moment of inertia was determined by the double pendulum method. The aircraft model was developed in such a way that the specific force and moment model (high bandwidth) is split from the point-mass dynamics of the aircraft (low bandwidth) [1]. The advantage of modelling the aircraft according to this unique method, results in a design that has simple decoupled linear controllers. The inner-loop controllers control the high-bandwidth specific accelerations and roll-rate, while the outer-loop controllers control the low-bandwidth point-mass dynamics. The performance of the developed auto-landing flight control system was tested in software-in-the-loop (SIL) and hardware-in-the-loop (HIL) simulations. A Monte Carlo non-linear landing simulation analysis showed that the FCS is expected to land the aircraft 95% of the time within a circle with a diameter of 1.5m. Practical flight tests verified the theoretical results of the developed controllers and the project was concluded with five autonomous landings. The aircraft landed within a circle with a 7.5m radius with the aiming point at the centre of the circle. In the practical landings the longitudinal landing error dominated the landing performance of the autonomous landing system. The large longitudinal error resulted from a climb rate bias on the estimated climb rate and a shallow landing glide slope. / AFRIKAANSE OPSOMMING: Hierdie skripsie stel die ontwikkeling en praktiese demonstrasie van ʼn self-landdende onbemande vastevlerkvliegtuigstelsel voor, wat op ʼn stilstaande platform te lande kan kom met behulp van ʼn uiters akkurate globale posisionering stelsel. Die projek maak deel uit van ʼn groter projek, waarvan die doel is om ʼn onbemande vastevlerkvliegtuig op ʼn bewegende platform te laat land (bv. op ʼn boot se dek) in onstuimige windtoestande. Die hoofdoel van die projek was om die vliegtuig so akkuraat as moontlik op die aanloopbaan te laat land. ʼn Vliegtuigraamwerk is vir dié doel gekies wat met gepaste avionica uitgerus is. Die uitgeruste vliegtuig se aerodinamsie eienskappe was geanaliseer met AVL en die traagheidsmoment is deur die dubbelependulum metode bepaal. Die vliegtuigmodel is op so ‘n manier onwikkel om [1] die spesifieke krag en momentmodel (vinnige reaksie) te skei van die puntmassadinamiek (stadige reaksie). Die voordeel van hierdie wyse van modulering is dat eenvoudige ontkoppelde beheerders ontwerp kon word. Die binnelusbeheerders beheer die vinnige reaksie-spesifieke versnellings en die rol tempo van die vliegtuig. Die buitelusbeheerders beheer die stadige reaksie puntmassa dinamiek. Die vliegbeheerstelsel is in sagteware-in-die-lus en hardeware-in-die-lus simulasies getoets. Die vliegtuig se landingseienskappe is ondersoek deur die uitvoer van Monte Carlo simulasies, die simulasie resultate wys dat die vliegtuig 95% van die tyd binne in ʼn sirkel met ʼn diameter van 1.5m geland het. Praktiese vlugtoetse het bevestig dat die teoretiese uitslae en die prakties uitslae ooreenstem. Die vliegtuig het twee suksesvolle outomatiese landings uitgevoer, waar dit binne ʼn 7.5m-radius sirkel geland het, waarvan die gewenste landingspunt die middelpunt was. In die outomatiese landings is die longitudinale landingsfout die grootse. Die groot longitudinale landingsfout is as gevolg van ʼn afset op die afgeskatte afwaartse spoed en ʼn lae landings gradiënt.

This thesis describes the design and implementation of systems for automatic landing and stabilization of air vehicle known as Quadrocopter. It focuses on image processing from a camera placed on board of the robot. The aim of image processing is detect landing target and calculate the distance from the desired position. Further is solved vertical stabilization for higher altitudes. For this function is measured atmospheric pressure. Conversely, for low altitude flight is used a ultrasonic range finder. In the conclusion is solved the implementation of the system.

Med målet att landa fastvingedrönare på en liten yta utfördes ett projekt som implementerade en ballistisk landning i simulation. Projektet utfördes på initiativ av Sjöräddningssällskapet som har utvecklat en fastvingedrönare som behöver kunna landa på en båt. Projektet var en del av ett kandidatarbete inom ämnet programvaruteknik med syfte attbesvara hur landningen kunde skapa värde för kunden, vilka erfarenheter för framtidensom projektgruppen kunde dokumentera och vilket stöd projektgruppen fick från att använda en systemanatomi. För att besvara frågeställningarna utarbetade projektgruppen en kravspecifikation och ensystemanatomi samt utförde projektet med kontinuerliga utvärderingar under projektetsgång. Vid utvecklingen användes en variant av Scrum. Som resultat implementerades engrundläggande version av en ballistisk landning. Dokument skapades för utvecklingenoch gemensamma erfarenheter samlades in genom utvärderingar och möten. Projektgruppen fick många lärdomar om att arbeta i och planera större projekt vilket diskuteras i rapporten. Projektet och den utvecklade landningens hållbarhetsaspekter utvärderades.

This thesis presents methods for robust precision landing of unmanned air vehicles (UAVs) on platforms at sea. Localization methods are proposed for UAV-to-boat state estimation for systems that employ real- time kinematic (RTK) global navigation satellite system (GNSS) and vision sensors. Solutions for GNSS-only are first presented, followed by the fusion of GNSS and vision. The important problem of sensor intrinsic calibration is solved with a novel offline batch estimation approach. Hardware results are presented for all methods. Our calibration of GNSS-to-camera is shown to estimate sensor offsets with millimeter level accuracy. Localization systems are combined with custom state machines that manage the landing attempt via a novel descent cone. This conical threshold enforces a safe and accurate landing. Our landing methods are demonstrated in real-world experiments and achieve consistent accurate landings with error below 10 cm. The fusion of camera and RTK is shown to produce a robust landing system with redundant localization sources.

<p>A control system for autonomous landing of an unmanned aerial vehicle (UAV)with high precision has been developed. The UAV is a medium sized model he-licopter. Measurements from a GPS, a camera and a compass are fused with anextended Kalman filter for state estimation of the helicopter. Four PID-controllers,one for each control signal of the helicopter, are used for the helicopter control.During the final test flights fifteen landings were performed with an average land-ing accuracy of 35 cm.    A bias in the GPS measurements makes it impossible to land the helicopter withhigh precision using only the GPS. Therefore, a vision system using a camera anda pattern provided landing platform has been developed. The vision system givesaccurate measurement of the 6-DOF pose of the helicopter relative the platform.These measurements are used to guide the helicopter to the landing target. Inorder to use the vision system in real time, fast image processing algorithms havebeen developed. The vision system can easily match up the with the camera framerate of 30 Hz.</p> / <p>Ett kontrolsystem för att autonomt landa en modellhelikopter har utvecklats.Mätdata från en GPS, en kamera samt en kompass fusioneras med ett Extend-ed Kalman Filter för tillståndsestimering av helikoptern. Fyra PID-regulatorer,en för varje kontrolsignal på helikoptern, har används för regleringen. Under densista provflygningen gjordes tre landingar av vilken den minst lyckade slutade35 cm från målet.    På grund av en drift i GPS-mätningarna är det omöjligt att landa helikopternmed hög precision med bara en GPS. Därför har ett bildbehandlingssystem som an-vänder en kamera samt ett mönster på platformen utvecklats. Bidbehandlingssys-temet mäter positionen och orienteringen av helikoptern relativt platformen. Dessamätningar används kompensera för GPS-mätningarnas drift. Snabba bildbehan-dlingsalgoritmer har utvecklats för att kunna använda bildbehandlingssystemet irealtid. Systemet är mycket snabbare än 30 bilder per sekund vilket är kameranshastighet.</p>

The thesis has been developed as part of the requirements for a PhD degree at the Artificial Intelligence and Integrated Computer System division (AIICS) in the Department of Computer and Information Sciences at Linköping University.The work focuses on issues related to Unmanned Aerial Vehicle (UAV) navigation, in particular in the areas of guidance and vision-based autonomous flight in situations of short and long term GPS outage.The thesis is divided into two parts. The first part presents a helicopter simulator and a path following control mode developed and implemented on an experimental helicopter platform. The second part presents an approach to the problem of vision-based state estimation for autonomous aerial platforms which makes use of geo-referenced images for localization purposes. The problem of vision-based landing is also addressed with emphasis on fusion between inertial sensors and video camera using an artificial landing pad as reference pattern. In the last chapter, a solution to a vision-based ground object geo-location problem using a fixed-wing micro aerial vehicle platform is presented.The helicopter guidance and vision-based navigation methods developed in the thesis have been implemented and tested in real flight-tests using a Yamaha Rmax helicopter. Extensive experimental flight-test results are presented. / WITAS

A control system for autonomous landing of an unmanned aerial vehicle (UAV)with high precision has been developed. The UAV is a medium sized model he-licopter. Measurements from a GPS, a camera and a compass are fused with anextended Kalman filter for state estimation of the helicopter. Four PID-controllers,one for each control signal of the helicopter, are used for the helicopter control.During the final test flights fifteen landings were performed with an average land-ing accuracy of 35 cm.    A bias in the GPS measurements makes it impossible to land the helicopter withhigh precision using only the GPS. Therefore, a vision system using a camera anda pattern provided landing platform has been developed. The vision system givesaccurate measurement of the 6-DOF pose of the helicopter relative the platform.These measurements are used to guide the helicopter to the landing target. Inorder to use the vision system in real time, fast image processing algorithms havebeen developed. The vision system can easily match up the with the camera framerate of 30 Hz. / Ett kontrolsystem för att autonomt landa en modellhelikopter har utvecklats.Mätdata från en GPS, en kamera samt en kompass fusioneras med ett Extend-ed Kalman Filter för tillståndsestimering av helikoptern. Fyra PID-regulatorer,en för varje kontrolsignal på helikoptern, har används för regleringen. Under densista provflygningen gjordes tre landingar av vilken den minst lyckade slutade35 cm från målet.    På grund av en drift i GPS-mätningarna är det omöjligt att landa helikopternmed hög precision med bara en GPS. Därför har ett bildbehandlingssystem som an-vänder en kamera samt ett mönster på platformen utvecklats. Bidbehandlingssys-temet mäter positionen och orienteringen av helikoptern relativt platformen. Dessamätningar används kompensera för GPS-mätningarnas drift. Snabba bildbehan-dlingsalgoritmer har utvecklats för att kunna använda bildbehandlingssystemet irealtid. Systemet är mycket snabbare än 30 bilder per sekund vilket är kameranshastighet.