Search

This paper proposes an adaptive path-following controller of small fixed-wing Unmanned Aerial Vehicles (UAVs) in the presence of wind disturbances, which explicitly considers that wind speed is time-varying. The main idea was to formulate UAVs path-following as control design for systems with parametric uncertainties and external disturbances. Assuming that there is no prior information on wind, the proposed solution is based on the L1 adaptive control, using linearized model dynamics. This approach makes clear statements for performance specifications of the controller and relaxes the common constant wind velocity assumption. This makes the design more realistic and the analysis more rigorous, because in practice wind is usually time varying (windshear, turbulence and gusting). The path-following controller was demonstrated in flight under wind speed up to 10m/s, representing 50% of the nominal UAV airspeed.

Like in other surveying works, UAV flights require prior work that involves flight planning and equipment preparation and, often, many complementary tasks. These may involve bringing together technicians from different domains, booking a car and possibly accommodation, and some time-consuming complementary bureaucratic work. Teams operating UAVs know how much the flights are affected by weather conditions. The wind is the weather variable that, in proportion (number of occurrences per year), causes the major number of changes to a scheduled work. Obtaining reliable information about the intensity of the wind, a few days in advance, is an asset for those who have to carry out the various tasks mentioned previously. There are several websites from which one can access weather forecasts. Is any website better because it presents more reliable data? The data and the analysis presented in the paper will give some clues. The data includes wind speed, registered daily, at 12:00 (pm) for a year, by a meteorological station with online data, which belongs to a meteorological institute. Also on a daily basis, several websites with meteorological data were consulted, and wind speed forecasts for the same hour for up to four days in advance were collected. An analysis of the data can provide information about whether there is a website that stands out for the quality of the forecasts, and if there is a need to consult several websites to have better information.

Multicopters such as quadcopters are a popular tool within industries such as mining, shipping and surveillance where a high level of autonomy can save time, increase efficiency and most importantly provide safety. While Unmanned Aerial Vehicles have been a big area in research and used in the mentioned industries, the level of autonomy is still low. Simple actions such as loading and offloading payload or swapping batteries is still a manual task performed by humans. If multicopters are to be used as an autonomous tool the need for solutions where the machines can perform the simplest task such as swapping batteries become an important stepping stone to reach the autonomy goals. Earlier works propose landing solutions focused on landing autonomous vehicles but the lack of accuracy is hindering the vehicles to safely dock with a landing platform. This thesis combines multiple areas such as trajectory generation, visual marker tracking and UAV control where results are shown in both simulation and laboratory experiments. With the use of a Model Predictive Controller for both trajectory generation and UAV control, a multicopter can safely land on a small enough platform which can be mounted on a small mobile robot. Additionally an algorithm to tune the trajectory generator is presented which shows how much weights can be increased in the MPC controller for the system to remain stable.

In order to operate safely in windy and gusty conditions, multirotor VTOL aircraft require gust resilience. This paper shows that their gust rejection properties can be improved by applying a small amount of fixed outward rotor tilt. Standard aerodynamic models of the rotors are incorporated into two dynamic models to assess the gust rejection properties. The first case is a conceptual birotor planar VTOL aircraft. The dependence of the trim and stability on the tilt angle are analyzed. The aircraft is stabilized using a pole-placement approach in order to obtain consistent closed-loop station-keeping performance in still air. The effect of gusts on the resulting response is determined by simulation. The second case study is for a quadrotor with a 10∘" role="presentation" style="max-height: none; display: inline; line-height: normal; word-spacing: normal; overflow-wrap: normal; white-space: nowrap; float: none; direction: ltr; max-width: none; min-width: 0px; min-height: 0px; border-width: 0px; border-style: initial; position: relative;">10∘ outward rotor tilt. The aerodynamic coefficients are analyzed for trimmed station-keeping over a range of steady wind speeds. An LQR controller is used to apply station-keeping that includes integral action, and the gust responses are again obtained using simulation. The results show that the outward rotor tilt causes the aircraft to pitch down into a lateral gust, providing lateral force that opposes the gust and so significantly improving the gust rejection properties.

Mobile crowdsensing (MCS) is attracting considerable attention in the past few years as a new paradigm for large-scale information sensing. Unmanned aerial vehicles (UAVs) have played a significant role in MCS tasks and served as crucial nodes in the newly-proposed space-air-ground integrated network (SAGIN). In this paper, we incorporate SAGIN into MCS task and present a Space-Air-Ground integrated Mobile CrowdSensing (SAG-MCS) problem. Based on multi-source observations from embedded sensors and satellites, an aerial UAV swarm is required to carry out energy-efficient data collection and recharging tasks. Up to date, few studies have explored such multi-task MCS problem with the cooperation of UAV swarm and satellites. To address this multi-agent problem, we propose a novel deep reinforcement learning (DRL) based method called Multi-Scale Soft Deep Recurrent Graph Network (ms-SDRGN). Our ms-SDRGN approach incorporates a multi-scale convolutional encoder to process multi-source raw observations for better feature exploitation. We also use a graph attention mechanism to model inter-UAV communications and aggregate extra neighboring information, and utilize a gated recurrent unit for long-term performance. In addition, a stochastic policy can be learned through a maximum-entropy method with an adjustable temperature parameter. Specifically, we design a heuristic reward function to encourage the agents to achieve global cooperation under partial observability. We train the model to convergence and conduct a series of case studies. Evaluation results show statistical significance and that ms-SDRGN outperforms three state-of-the-art DRL baselines in SAG-MCS. Compared with the best-performing baseline, ms-SDRGN improves 29.0% reward and 3.8% CFE score. We also investigate the scalability and robustness of ms-SDRGN towards DRL environments with diverse observation scales or demanding communication conditions.

Modeling and controlling highly nonlinear, multivariable, unstable, coupled and underactuated systems are challenging problems to which a unique solution does not exist. Modeling and control of Unmanned Aerial Vehicles (UAVs) with four rotors fall into that category of problems. In this paper, a nonlinear quadrotor UAV dynamical model is developed with the Newton–Euler method, and a control architecture is proposed for 3D trajectory tracking. The controller design is decoupled into two parts: an inner loop for attitude stabilization and an outer loop for trajectory tracking. A few attitude stabilization methods are discussed, implemented and compared, considering the following control approaches: Proportional–Integral–Derivative (PID), Linear–Quadratic Regulator (LQR), Model Predictive Control (MPC), Feedback Linearization (FL) and Sliding Mode Control (SMC). This paper is intended to serve as a guideline work for selecting quadcopters’ control strategies, both in terms of quantitative and qualitative considerations. PID and LQR controllers are designed, exploiting the model linearized about the hovering condition, while MPC, FL and SMC directly exploit the nonlinear model, with minor simplifications. The fast dynamics ensured by the SMC-based controller together with its robustness and the limited estimated command effort of the controller make it the most promising controller for quadrotor attitude stabilization. The outer loop consists of three independent PID controllers: one for altitude control and the other two, together with a dynamics’ inversion, are entitled to the computation of the reference attitude for the inner loop. The capability of the controlled closed-loop system of executing complex trajectories is demonstrated by means of simulations in MATLAB/Simulink®.

This paper addresses the problem of robustly controlling an actively tilting quadrotor UAV. The proposed technique is model-free and it is based on hyperbolic functions of the six-dimensional pose error of the UAV with respect to the world reference frame; this hyperbolic controller globally attracts the error signals to an ultimate bound about the origin despite external disturbances, which is proved by way of a strict Lyapunov function based analysis. The effectiveness of the controller is evaluated by means of tracking and regulation experiments on adverse conditions, which were implemented on a virtual model of the UAV through a physics-engine-based simulation environment that provides an almost identical behaviour than a real UAV. The norm of the six-dimensional error signal converged to zero for the regulation experiments, whereas for tracking it did not exceed 0.05 meters, which indicated a successful operation of the control system. In addition, the performance of the hyperbolic controller was contrasted against a nonlinear PID, which resulted in a better performance in favour of the first one, who settled the errors to zero up to eight seconds before and demanded up to 2000 less revolutions per minute from the rotors while performing the same regulation tasks. All the aforesaid successful results place the proposed technique as a competitive alternative for controlling actively tilting multirotors due to its simplicity, robustness and demonstrated effectiveness.

Absolute localization of a flying UAV on its own in a global-navigation-satellite-system (GNSS)-denied environment is always a challenge. In this paper, we present a landmark-based approach where a UAV is automatically locked into the landmark scene shown in a georeferenced image via a feedback control loop, which is driven by the output of an aerial image registration. To pursue a real-time application, we design and implement a speeded-up-robust-features (SURF)-based image registration algorithm that focuses efficiency and robustness under a 2D geometric transformation. A linear UAV controller with signals of four degrees of freedom is derived from the estimated transformation matrix. The approach is validated in a virtual simulation environment, with experimental results demonstrating the effectiveness and robustness of the proposed UAV self-localization system.

[EN] Guidance, navigation, and control system design is, undoubtedly, one of the most relevant issues in any type of unmanned aerial vehicle, especially in the case of military missions. This task needs to be performed in the most efficient way possible, which involves trying to satisfy a set of requirements that are sometimes in opposition. The purpose of this article was to compare two different control strategies in conjunction with a path-planning and guidance system with the objective of completing military missions in the most satisfactory way. For this purpose, a novel dynamic trajectory-planning algorithm is employed, which can obtain an appropriate trajectory by analyzing the environment as a discrete 3D adaptive mesh and performs a softening process a posteriori. Moreover, two multivariable control techniques are proposed, i.e., the linear quadratic regulator and the model predictive control, which were designed to offer optimal responses in terms of stability and robustness., This work was partially funded by project RTI2018-096904-B-I00 from the Spanish Ministry of Economy and by project AICO/2019/055 from Generalitat Valenciana.

Обсяг пояснювальної записки магістерської дисертації становить 108 сторінок, які включають в себе 7 розділів, 27 ілюстрацій, 37 таблиць, 29 бібліографічних найменувань за переліком джерел посилань. Актуальність теми дослідження. Завдяки новим регламентам ЄС та запуску космічної програми, кожен безпілотний літальний апарат (БПЛА) відтепер повинен транслювати телеметричні дані та ідентифікаційні дані на сервер, щоб забезпечити операції високої щільності з декількома автоматизованими БПЛА. Отже, вирішено створити стабільний багатоканальний зв’язок з безпілотником, для забезпечення можливістю передачі даних в обидва боки на сервер без будь-яких втрат інформації на великих відстанях. Мета та завдання. Створення стабільного багатоканального зв’язку сервера з безпілотником, для передачі даних в обидва боки на сервер без будь-яких втрат інформації на великих відстанях. Об’єктом дослідження. Процес передачі даних з БПЛА до сервера за допомогою інтернет-з’єднання. Предмет дослідження. Методи передачі даних з БПЛА до сервера. Методи дослідження. В роботі використовувалися методи теорії ймовірностей, лінійної і нелінійної фільтрації в дискретному часі, статистичної теорії прийняття рішень і статистичного моделювання на ЕОМ, графічні методи визначення даних. Наукова новизна одержаних результатів: Розроблено систему передачі даних для безпілотного літального апарата для визначення його місцезнаходження, керування польотом та синхронізації з іншими БПЛА для уникнення зіткнень в цій системі на відмну від існуючих передача та отримання даних відбувається відразу з БПЛА на сервер через інтернетмережу, а не через пульт керування, що в свою чергу дозволяє здійснювати, відслідковувати та корегувати польоти з будь-якої точки планети, що дозволяє експлуатувати БПЛА без задіяння пілотів у безпосередній близькості до нього. Практичне значення одержаних результатів: 1. Одержані результати показують можливість польотів БПЛА над досліджуваною територією без втрати даних, якими обмінюються літальний апарат та сервер за допомогою мережі інтернет 2. Конфігурація з трьох систем зв’язку та окремого контролера забезпечує можливість сталого обміну інформацією між БПЛА та сервером з достатньою швидкість інтернет-з’єднання, що дозволяє передавати телеметричні дані, дані для керування дроном та відео з польотної камери. 3. Розроблений алгоритм програми використання модулів зв'язку не вимагає значних обчислювальних витрат і може бути реалізовано на сучасних мікропроцесорах в масштабі реального часу. 4. Використання готових технологій передачі даних (LTE, Wi-Fi, LoRa) знижує вартість системи без втрати надійності з’єднання Апробація результатів дисертації. Матеріали дисертації та основні положення, отримані в роботі, були використані у ході наукової діяльності у Пражськом технічному університеті, м. Прага, Чехія; The volume of the explanatory note of the master's thesis is 108 pages, which include 7 sections, 27 illustrations, 37 tables, 29 bibliographic names according to the list of sources of links. Relevance of the research topic. hanks to the new EU regulations and the launch of the space program, every unmanned aerial vehicle (UAV) must now transmit telemetry and identification data to the server to provide high-density operations with several automated UAVs. Therefore, it is decided to create a stable multi-channel communication with the drone, in order to enable data transfer on both sides to the server without any loss of information over long distances. Purpose and tasks. Creating a stable multi-channel communication server with the drone, for data transfer on both sides to the server without any loss of information over long distances. The object of study. The process of transmitting data from a UAV to a server via an Internet connection. Object of research. The object is the process of determining the location of RFID tags. Subject of research. Methods of data transfer from UAV to server. Research methods. The methods of the theory of probabilities, linear and nonlinear filtering in discrete time, statistical decision theory and statistical modeling on computers, graphical methods of data determination were used in the work. Research methods. The methods of the theory of probabilities, linear and nonlinear filtering in discrete time, statistical decision theory and statistical modeling on computers, graphical methods of data determination were used in the work. Scientific novelty of the obtained results. A data transfer system for an unmanned aerial vehicle was developed to determine its mapping, flight control and synchronization with other UAVs in order to avoid collisions in this system, as opposed to existing transmissions and receiving data from the UAV at server through the Internet, not through the control panel, which in turn allows you to perform, track and adjust flights from anywhere on the planet, which allows UAVs to operate without piloting into the no-holds-barred some closeness to it. The practical significance of the results obtained. The practical significance of the results obtained: 1. The results obtained show the possibility of UAV flights over the investigated territory without loss of data exchanged by the aircraft and server via the Internet. 2. The configuration of the three communication systems and the individual controller ensures the constant exchange of information between the UAV and the server with sufficient internet connection speed to transmit telemetry data, drone control data and video from the flight camera. 3. The developed algorithm of the program of use of communication modules does not require considerable computational cost and can be implemented on modern microprocessors in real time scale. 4. The use of ready-made data transfer technologies (LTE, Wi-Fi, LoRa) reduces the cost of the system without losing connection reliability Testing the results of the thesis. The dissertation materials and the main provisions obtained in the work were used during the scientific activity at the Prague Technical University (Prague, Czech Republic 2018).

National audience; Teleoperation of Unmanned Aerial Vehicles (UAVs) has recently become an noteworthly research topic in the field of human robot interaction. Each year, a variety of devices is being studied to design adapted interface for diverse purpose such as view taking, search and rescue operation or suveillance. New interfaces have to be precise, simple and intuitive even for complex path planning. Moreover, when teleoperation involves long distance control, user needs to get proper feedbacks and avoid motion sickness. In order to overcome all these challenges, a new interaction metaphor named DrEAM (Drone Exocentric Advanced Metaphor) was designed. User can see the UAV he is controlling in a virtual environment mapped to the real world. He can interact with it as a simple object in a classical virtual world. An experiment was lead in order to evaluate the perfomances of this metaphor, comparing performance of novice user using either a direct-view joystick control or using DrEAM.

11th International Conference on Electronics, Computers and Artificial Intelligence, ECAI (2019 ; Pitesti, Romania), Recent developments in sensor technology, actuators and energy storage devices provides opportunities for development of micro flying robots. This paper provides information about modeling and control of a Quadrotor. Proportional Integral Derivative (PID) controller is designed for altitude and attitude control of Quadrotor. Various simulations were performed and it could be interpreted that the PID controller is capable of tracking the desired reference values. Finally, the last section is devoted for discussing the results of each simulation. © 2019 IEEE.

Current UAV-recorded datasets are mostly limited to action recognition and object tracking, whereas the gesture signals datasets were mostly recorded in indoor spaces. Currently, there is no outdoor recorded public video dataset for UAV commanding signals. Gesture signals can be effectively used with UAVs by leveraging the UAVs visual sensors and operational simplicity. To fill this gap and enable research in wider application areas, we present a UAV gesture signals dataset recorded in an outdoor setting. We selected 13 gestures suitable for basic UAV navigation and command from general aircraft handling and helicopter handling signals. We provide 119 high-definition video clips consisting of 37151 frames. The overall baseline gesture recognition performance computed using Pose-based Convolutional Neural Network (P-CNN) is 91.9%. All the frames are annotated with body joints and gesture classes in order to extend the dataset’s applicability to a wider research area including gesture recognition, action recognition, human pose recognition and situation awareness. Refereed/Peer-reviewed

International audience; In this paper, a constructive procedure to design Functional Unknown Input Observers for nonlinear continuous time systems is proposed under the Polytopic Takagi-Sugeno framework. An equivalent representation for the nonlinear model is achieved using the sector nonlinearity transformation. Applying the Lyapunov theory and the L2 attenuation, linear matrix inequalities conditions are deduced which are solved for feasibility to obtain the observer design matrices. To cope with the effect of unknown inputs, classical approach of decoupling the unknown input for the linear case is used. Both algebraic and solver-based solutions are proposed (relaxed conditions). Necessary and sufficient conditions for the existence of the functional polytopic observer are given. For both approaches, the general and particular cases (measurable premise variables, full state estimation with full and reduced order cases) are considered and it is shown that the proposed conditions correspond to the one presented for standard linear case. To illustrate the proposed theoretical results, detailed numerical simulations are presented for a Quadrotor Aerial Robots Landing and a Waste Water Treatment Plant. Both systems are highly nonlinear and represented in a T-S polytopic form with unmeasurable premise variables and unknown inputs.

Purpose – The growing use of small unmanned rotorcraft in civilian applications means that safe operation is increasingly important. The purpose of this paper is to investigate the fault tolerant properties to faults in the actuators of an C 1 adaptive controller for a quadrotor vehicle. Design/methodology/approach – C 1 adaptive control provides fast adaptation along with decoupling between adaptation and robustness. This makes the approach a suitable candidate for fault tolerant control of quadrotor and other multirotor vehicles. In the paper, the design of an C 1 adaptive controller is presented. The controller is compared to a fixed-gain LQR controller. Findings – The C 1 adaptive controller is shown to have improved performance when subject to actuator faults, and a higher range of actuator fault tolerance. Research limitations/implications – The control scheme is tested in simulation of a simple model that ignores aerodynamic and gyroscopic effects. Hence for further work, testing with a more complete model is recommended followed by implementation on an actual platform and flight test. The effect of sensor noise should also be considered along with investigation into the influence of wind disturbances and tolerance to sensor failures. Furthermore, quadrotors cannot tolerate total failure of a rotor without loss of control of one of the degrees of freedom, this aspect requires further investigation. Practical implications – Applying the C 1 adaptive controller to a hexrotor or octorotor would increase the reliability of such vehicles without recourse to methods that require fault detection schemes and control reallocation as well as providing tolerance to a total loss of a rotor. Social implications – In order for quadrotors and other similar unmanned air vehicles to undertake many proposed roles, a high level of safety is required. Hence the controllers should be fault tolerant. Originality/value – Fault tolerance to partial actuator/effector faults is demonstrated using an C 1 adaptive controller.

The effect of rotor aerodynamics on the stability and control of multi-rotor aircraft is explored by analysis of a conceptual planar birotor aircraft. In particular, the effect of rotor tilt on the stability and zero-location is investigated. Furthermore, it is shown that designing the vehicle with outwards rotor tilt potentially results in dramatic improvement in the gust rejection properties of the vehicle.

An enhanced level of autonomy for UAVs is desirable when behaviours more sophisticated than simply following waypoint presets are desired. Behaviour Trees (BT) provide a decision tree like approach for developers to efficiently implement these sophisticated behaviours through a robust, modular, scalable framework. This paper presents a framework for easy development of BTs which implements autonomous unmanned aerial vehicle (UAV) behaviours, building upon the existing Java Behaviour Tree (JBT) software for video games. Behaviour Engine software was also developed, allowing the BTs to be interfaced with an autopilot to deliver enhanced UAV control. The Behaviour Engine software was loaded with BTs implementing the UAV behaviours of pursue, evade, stalk and patrol, with the results showing that these behaviours function as desired in the Behaviour Engine software for two simulated UAVs.

In the coming years, unmanned combat air vehicles will be designed to outperform most pilot-in-the-loop systems. The absence of a pilot onboard allows larger flight envelopes to be considered with, for example, higher load factors. From a control perspective, the above induces new needs for improved design methods to obtain robust controllers that will automatically adapt to extremely varying flight conditions. Based on the well-known concept of nonlinear dynamic inversion (NDI), the approach of this paper introduces an original and rather generic “robustification” framework, which leads to a multi-objective H∞ design problem. The latter is now easily solved by existing and efficient numerical tools based on nonsmooth optimization. The proposed design methodology is illustrated by a combat aircraft control problem., AerospaceLab Journal, Issue 8, December 2014; ISSN: 2107-6596

International audience; In this paper, we present two control strategies of a quadrotor Unmanned Aerial Vehicle (UAV). The dynamics of the system is controlled to achieve a complex trajectory tracking. A controller using the integral backstepping is developed for the stabilization of the six degrees of freedom (6DoF) of the system. Furthermore, we propose a hybrid strategy by combining the integral backstepping and the PID linear controller. For this, we divide the dynamic system into two subsystems; each one with 3DoF. For the stabilization of the rotational dynamics, the PID controller is proposed. The Integral Backstepping is applied for the longitudinal, lateral and vertical position stabilization. Finally, some numerical simulations are conducted to illustrate the robustness of both strategies.

For a robot to autonomously localise or position itself with respect to its environ- ment, a key requirement is the perception of this environment. In this respect, the visual information provided by a camera is a particularly rich source of information, commonly used in robotics. Our work deals with the use of visual information in the context of UAV control. More speciﰀcally, two tasks have been considered : ﰀrst, a tracking task, in which a UAV has to follow a moving object - a car - in an unknown environment, second, a positioning task for a UAV navigating in a structured GPS-deprived environment. In both cases, we propose full approaches, considering both the robust extraction of appropriate visual informations and the visual servoing of the UAV. The experi- ments performed on a small quadrotor UAV show the validity of our approaches.; Pour qu'un système robotisé puisse accomplir de façon autonome des fonctions en apparence simples, telles que se localiser ou se positionner par rapport à son environnement, il doit avant tout percevoir cet environnement. La perception vi- suelle obtenue à l'aide d'une caméra constitue à cet égard une source d'information particulièrement riche, largement utilisée en robotique. Le travail présenté dans cette thèse concerne l'usage d'informations visuelles dans le contexte de la commande de mini-drones. En particulier deux types de tâches ont été considérées : une tâche de poursuite, dans laquelle un objet - une voiture - se déplace dans un environnement inconnu et l'on souhaite qu'un drone puisse suivre son mouvement, et une tâche de positionnement ou de navigation pour un drone évoluant dans un environnement structuré - intérieur de bâtiment - dans lequel le signal GPS n'est pas disponible. Dans les deux cas, nous avons proposé des approches complètes, depuis l'extrac- tion robuste d'informations visuelles jusqu'à la commande d'un drone à partir de ces informations. Des expériences mises en ÷uvre sur un mini-drone quadrirotor montrent la validité des approches proposées. Mots-clefs : Vision par ordinateur, asservissement visuel, commande de drone

Most applications of Unmanned Aerial Vehicles are related to events that occur on the ground. In particular, ground target tracking, be the target static, slowly moving or maneuvering at high speeds, is an essential task for UAVs. The overall objective of this thesis is to provide methods to endow a drone to autonomously track a moving ground target, under the following conditions: - A fixed wing UAV equipped with a monocular camera. - Presence of obstacles that hinder ground visibility. - No Fly Zones that limit the airspace. - Restrictions on the field of view of the observing sensor (a camera) - Various target dynamics and behavior: the target may be either moving on an open field or on a road network, and also has dynamic constraints (e.g. if it is a car). It can be neutral or evasive: in the latter case, it can exploit the presence of obstacles, denoted as "shadows" to avoid being tracked by the UAV, making the problem akin to a "hide and seek" game. The thesis proposes three approaches to tackle this problem: - A control based navigation method, - An adversarial predictive method, - And a discrete game theoretic approach. Results obtained in realistic simulations and with an actual UAV are presented to evaluate and compare the pros and cons of each approach. Extensions to the multi-drone case are also considered.; La plupart des applications des avions drones sont liées à l'observation d'événements au sol. En particulier, les suivi de cibles terrestres mobiles, qu'elles soient statiques, lentes ou rapides, est une tâche essentielle pour un drone. L'objectif global de la thèse est de proposer des méthodes qui permettent à un drone de suivre une cible terrestre, dans les conditions suivantes: - Le drone est de type voilure fixe équipé d'une caméra monoculaire. - Présence d'obstacles qui occultent la visibilité de zones au sol. - Existence de zones d'exclusion aérienne qui limitent le mouvement aérien. - Restrictions sur le champ de vue du capteur qui assure le suivi (caméra) - Différents comportements de la cible : elle peut évoluer librement ou sous contraintes dynamiques (cas d'une voiture par exemple), et peut être neutre ou évasive~: dans ce dernier cas, elle peut exploiter la présence d'obstacles pour éviter d'être perçue par le drone. Trois approches pour aborder ce problème sont proposées dans la thèse : - Une méthode basée aux lois de contrôle et de la navigation, - Une méthode basée sur la prédiction des déplacements de la cible, - Et une approche basée sur la théorie des jeux. Des résultats obtenus par des simulations réalistes et avec un drone sont présentés, pour évaluer et comparer les avantages et inconvénients de chacune des approches. Des extensions au cas "multi-drones" sont aussi proposées.

Most applications of Unmanned Aerial Vehicles are related to events that occur on the ground. In particular, ground target tracking, be the target static, slowly moving or maneuvering at high speeds, is an essential task for UAVs. The overall objective of this thesis is to provide methods to endow a drone to autonomously track a moving ground target, under the following conditions: - A fixed wing UAV equipped with a monocular camera. - Presence of obstacles that hinder ground visibility. - No Fly Zones that limit the airspace. - Restrictions on the field of view of the observing sensor (a camera) - Various target dynamics and behavior: the target may be either moving on an open field or on a road network, and also has dynamic constraints (e.g. if it is a car). It can be neutral or evasive: in the latter case, it can exploit the presence of obstacles, denoted as "shadows" to avoid being tracked by the UAV, making the problem akin to a "hide and seek" game. The thesis proposes three approaches to tackle this problem: - A control based navigation method, - An adversarial predictive method, - And a discrete game theoretic approach. Results obtained in realistic simulations and with an actual UAV are presented to evaluate and compare the pros and cons of each approach. Extensions to the multi-drone case are also considered.; La plupart des applications des avions drones sont liées à l'observation d'événements au sol. En particulier, les suivi de cibles terrestres mobiles, qu'elles soient statiques, lentes ou rapides, est une tâche essentielle pour un drone. L'objectif global de la thèse est de proposer des méthodes qui permettent à un drone de suivre une cible terrestre, dans les conditions suivantes: - Le drone est de type voilure fixe équipé d'une caméra monoculaire. - Présence d'obstacles qui occultent la visibilité de zones au sol. - Existence de zones d'exclusion aérienne qui limitent le mouvement aérien. - Restrictions sur le champ de vue du capteur qui assure le suivi (caméra) - Différents comportements de la cible : elle peut évoluer librement ou sous contraintes dynamiques (cas d'une voiture par exemple), et peut être neutre ou évasive~: dans ce dernier cas, elle peut exploiter la présence d'obstacles pour éviter d'être perçue par le drone. Trois approches pour aborder ce problème sont proposées dans la thèse : - Une méthode basée aux lois de contrôle et de la navigation, - Une méthode basée sur la prédiction des déplacements de la cible, - Et une approche basée sur la théorie des jeux. Des résultats obtenus par des simulations réalistes et avec un drone sont présentés, pour évaluer et comparer les avantages et inconvénients de chacune des approches. Des extensions au cas "multi-drones" sont aussi proposées.

In this brief, we consider the problem of 3-D path generation and tracking for unmanned air vehicles (UAVs). The proposed path generation algorithm allows us to find a path satisfying arbitrary initial and final conditions, specified in terms of position and velocity. Our method assumes that aircraft structural and dynamic limitations can be translated in a turn radius constraint; therefore, the generated paths satisfy a constraint on the minimum admissible turning radius. The proposed algorithm for the path tracking guarantees, under specified assumptions, that the tracking error, both in position and in attitude, asymptotically tends to zero. The work has been carried out with reference to the UAV of the Italian Aerospace Research Center (CIRA). Simulation results for both the path generation and the tracking algorithms are presented; the latter have been obtained using a detailed 6-degree-of-freedom model of the CIRA UAV in the presence of wind and turbulence.