This work focuses on the problem that pertains to area searching with UAVs. Specifically developing algorithms that generate flight paths that are short with- out sacrificing flyability. For instance, very sharp turns will compromise flyability since fixed wing aircraft cannot make very sharp turns. This thesis provides an analysis of different types of search methods, area decompositions, and combi- nations thereof. The search methods used are side to side searching and spiral searching. In side to side searching the aircraft goes back and forth only making 90-degree turns. Spiral search searches the shape in a spiral pattern starting on the outer perimeter working its way in. The idea being that it should generate flight paths that are easy to fly since all turns should be with a large turn radii. Area decomposition is done to divide complex shapes into smaller more manage- able shapes. The report concludes that with the implemented methods the side to side scanning method without area decomposition yields good and above all very reliable results. The reliability stems from the fact that all turns are 90 degrees and that algorithm never get stuck or makes bad mistakes. Only having 90 degree turns results in only four different types of turns. This allows the airplanes behav- ior along the route to be predictable after flying the first four turns. Although this assumes that the strength of the wind is a greater influence than the turbulences effect on the aircraft’s flight characteristics. This is a very valuable feature for an operator in charge of a flight. The other tested methods and area decompositions often yield a shorter flight path, however, despite extensive adjustments to the algorithms they never came to handle all cases in a satisfactory manner. These methods may also generate any kind of turn at any time, including turns of nearly 180 degrees. These turns can lead to an airplane missing the intended flight path and thus missing to scan the intended area properly. Area decomposition proves to be really effective only when the area has many protrusions that stick out in different directions, think of a starfish shape. In these cases the side to side algo- rithm generate a path that has long legs over parts that are not in the search area. When the area is decomposed the algorithm starts with, for example, one arm of the starfish at a time and then search the rest of the arms and body in turn.

In a distributed real-time environment where it is imperative to make correct decisions it is important to have all facts available to make the most accurate decision in a certain situation. An example of such an environment is an Unmanned Aerial Vehicle (UAV) system where several UAVs cooperate to carry out a certain task and the data recorded is analyzed after the completion of the mission. This project aims to define and implement a time series architecture for use together with a distributed real-time database for the ability to store temporal data. The result from this project is a time series (TS) architecture that uses DeeDS, a distributed real-time database, for storage. The TS architecture is used by an application modelled from a UAV scenario for storing temporal data. The temporal data is produced by a simulator. The TS architecture solves the problem of storing temporal data for applications using DeeDS. The TS architecture is also useful as a foundation for integrating time series in DeeDS since it is designed for space efficiency and real-time requirements.

After a huge amount of success within the military, the benefits of the use of unmanned aerial systems over manned aircraft is obvious. They are becoming cheaper and their functions advancing to such a point that there is now a large drive for their use by civilian operators. However there are a number of significant challenges that are slowing their inevitable integration into the national airspace systems of countries. A large array of emergency situations will need to be dealt with autonomously by contingency management systems to prevent potentially deadly incidences. One such emergency situation that will need autonomous intervention, is the total loss of thrust from engine failure. The complex multi faceted task of landing the stricken aircraft at a potentially unprepared site is called a forced landing. This thesis presents methods to address a number of critical parts of a forced landing system for use by an unmanned aerial system. In order for an emergency landing site to be considered, it needs to be within glide range. In order to find a landing site s reachability from the point of engine failure the aircraft s glide performance and a glide path must be known. A method by which to calculate the glide performance, both from aircraft parameters or experiments is shown. These are based on a number of steady state assumptions to make them generic and quick to compute. Despite the assumptions, these are shown to have reasonable accuracy. A minimum height loss path to the landing site is defined, which takes account of a steady uniform wind. While this path is not the path to be flown it enables a measure of how reachable a landing site is, as any extra height the aircraft has once it gets to the site makes a site more reachable. It is shown that this method is fast enough to be run online and is generic enough for use on a range of aircraft. Based on identified factors that make a landing site more suitable, a multi criteria decision making Bayesian network is developed to decide upon which site a unmanned aircraft should land in. It can handle uncertainty and non-complete information while guaranteeing a fast reasonable decision, which is critical in this time sensitive situation. A high fidelity simulation environment and flight test platform are developed in order to test the performance of the developed algorithms. The test environments developed enable rapid prototyping of algorithms not just within the scope of this thesis, but on a range of vehicle types. In simulation the minimum height loss paths show good accuracy, for two completely different types of aircraft. The decision making algorithms show that they are capable of being ran online in a flight test. They make a reasonable decision and are capable of quickly reacting to changing conditions, enabling redirection to a more suitable landing site.

Unmanned Aerial Vehicles (UAV) have been heavily studied in the past decade, where autonomous flights have been a popular subject. More complex applications have led to higher requirements on the autonomous flight algorithms and the absence of performance data complicates the selection of what algorithm to use for various applications. Therefore, this thesis focused in analyzing the performance difference between two methods, Simultaneous Localization AndMapping (SLAM) and Artificial Potential Field Approach (APFA), which are planning and reactive algorithms, respectively. Fundamental dynamics were applied, Feedback Linear Controllers (FBLC)s for stabilization and an odometry position model combined with an inverse dynamics technique that linearizes the non-linear odometry model. The SLAM approach was set up in four steps: landmark extraction which uses a point distance based method for segment separation, combined with a Split-And-Merge algorithm for extracting linear landmarks, data association that validates the landmarks, Extended Kalman Filter (EKF) that uses the landmarks together with the odometry model for estimating the position of the UAV, and a modified TangentBug as the reactive algorithm. The APFA was constructed of two functions, an attractive and a repulsive function. The two methods were implemented on the robotics simulation platform Virtual Robot Experimentation Platform (V-REP), where a quadcopter was used as the model for the UAV. All theory was implemented onto the quadcopter model and embedded scripts were used for communication within V-REP, mainly through internal Application Programming Interface (API)-functions. Furthermore, a script was written that randomly generates three different types of simulation environments. The implementation of both methods was analyzed in reaching an arbitrary goal position in terms of: the most successful, the most time efficient and the safest navigation path. Another thing analyzed was the time- and space-complexity of both implemented methods. The results stated that the implemented APFA and the SLAM approach had approximately equal success rate, SLAM had the safest navigation, was the most time efficient, and had the highest time- and space-complexity for a worst case scenario. One of the conclusions were that improvements could be done in the implementations. Future work includes adding a proper damping method, improving the flaws in the implemented methods as well as to use V-REP as a Robot Operating System (ROS)-node for creating a Software In The Loop (SITL)-simulation, in order to achieve more realistic simulations.

Unmanned aerial vehicles (UAV) have become an interesting and active research topic for photogrammetry. Current research is based on images acquired by an UAV, which have a high ground resolution and good spectral and radiometrical resolution, due to the low flight altitudes combined with a high resolution camera. UAV image flights are also cost effective and have become attractive for many applications including, change detection in small scale areas. One of the main problems preventing full automation of data processing of UAV imagery is the degradation effect of blur caused by camera movement during image acquisition. This can be caused by the normal flight movement of the UAV as well as strong winds, turbulence or sudden operator inputs. This blur disturbs the visual analysis and interpretation of the data, causes errors and can degrade the accuracy in automatic photogrammetric processing algorithms. The detection and removal of these images is currently achieved manually, which is both time consuming and prone to error, particularly for large image-sets. To increase the quality of data processing an automated process is necessary, which must be both reliable and quick. This thesis proves the negative affect that blurred images have on photogrammetric processing. It shows that small amounts of blur do have serious impacts on target detection and that it slows down processing speed due to the requirement of human intervention. Larger blur can make an image completely unusable and needs to be excluded from processing. To exclude images out of large image datasets an algorithm was developed. The newly developed method makes it possible to detect blur caused by linear camera displacement. The method is based on human detection of blur. Humans detect blurred images best by comparing it to other images in order to establish whether an image is blurred or not. The developed algorithm simulates this procedure by creating an image for comparison using image processing. Creating internally a comparable image makes the method independent of additional images. However, the calculated blur value named SIEDS (saturation image edge difference standard-deviation) on its own does not provide an absolute number to judge if an image is blurred or not. To achieve a reliable judgement of image sharpness the SIEDS value has to be compared to other SIEDS values of the same dataset. This algorithm enables the exclusion of blurred images and subsequently allows photogrammetric processing without them. However, it is also possible to use deblurring techniques to restor blurred images. Deblurring of images is a widely researched topic and often based on the Wiener or Richardson-Lucy deconvolution, which require precise knowledge of both the blur path and extent. Even with knowledge about the blur kernel, the correction causes errors such as ringing, and the deblurred image appears muddy and not completely sharp. In the study reported in this paper, overlapping images are used to support the deblurring process. An algorithm based on the Fourier transformation is presented. This works well in flat areas, but the need for geometrically correct sharp images for deblurring may limit the application. Another method to enhance the image is the unsharp mask method, which improves images significantly and makes photogrammetric processing more successful. However, deblurring of images needs to focus on geometric correct deblurring to assure geometric correct measurements. Furthermore, a novel edge shifting approach was developed which aims to do geometrically correct deblurring. The idea of edge shifting appears to be promising but requires more advanced programming.

Syftet med denna rapport var att utvärdera hur det på ett enkelt sätt och med hög tillförlitlighet går att kontrollera flygmätningar med UAV. Frågeställningarna som undersöktes var: Hur många kontrollpunkter krävs det för att kunna uppvisa en viss kvalité vid mätning i flygbilder tagna med UAV? Hur förändras kvalitén beroende på antalet kontrollpunkter? Hur ska en marksignal se ut för bästa synlighet i en bild? I den första delen av studien diskuterades olika varianter på hur en kontrollpunkt kan se ut. Sex stycken olika varianter tillverkades och testades på gamla flygfältet i Karlstad. Dessa granskades sedan okulärt, de två bästa valdes ut som sedan undersöktes i Global Mapper och beräknades i Excel. Den bästa kontrollpunkten erhölls och ytterligare signaler av den utformningen tillverkades, totalt 21 stycken. Den valda utformningen provades och flögs över en bergtäkt i Sandseryd utanför Jönköping. Data från flygningen beräknades och ett resultat erhölls. Resultatet visade på att man med ett par väl definierade kontrollpunkter, kan uppnå ett värde på under 25 mm radiellt i plan, ett medelfel under 10 mm i plan och drygt 15 mm i höjd. Det räcker således med ett par punkter men ett par extra rekommenderas, två punkter kan lika väl användas som tjugo punkter. En marksignal ska för bästa synlighet ha ett schackrutigt mönster i svart och vitt.

This Master´s Thesis Usage of drones in the Logistics of the company SKODA AUTO a.s. reacts to the current trend of unmanned aerial vehicles (drones) which are being used in many areas. Aim of this thesis is to analyse laws which regulate the operation of drones in selected countries, to assess suitability of drones for specific process of inventory of empty containers in the company SKODA AUTO a.s. and to suggest an alternative solution for this process. Thesis investigates usage of drones in the Logistics mainly for goods delivery and stock inventory. Mass drone delivery is currently not feasible due to many problems and challenges. Stock inventory has become a reality. Analysis of legislation is focused on EU, Czech Republic, Germany and China. Technology for inventory of empty containers using drones in SKODA AUTO a.s. has a long payback period (21 years) and author does not recommend drones for this process based on this criterion. As an alternative solution, the technology designed at AUDI company is recommended (camera placed on forklift and software for automated container identification).

La miniaturisation des calculateurs a permis le développement des drones, engins volants capable de se déplacer de façon autonome et de rendre des services, comme se rendre clans des lieux peu accessibles ou remplacer l'homme dans des missions pénibles. Un enjeu essentiel dans ce cadre est celui de l'information qu'ils doivent utiliser pour se déplacer, et donc des capteurs à exploiter pour obtenir cette information. Or nombre de ces capteurs présentent des inconvénients (risques de brouillage ou de masquage en particulier). L'utilisation d'une caméra vidéo dans ce contexte offre une perspective intéressante. L'objet de cette thèse était l'étude de l'utilisation d'une telle caméra dans un contexte capteur minimaliste: essentiellement l'utilisation des données visuelles et inertielles. Elle a porté sur le développement de lois de commande offrant au système ainsi bouclé des propriétés de stabilité et de robustesse. En particulier, une des difficultés majeures abordées vient de la connaissance très limitée de l'environnement dans lequel le drone évolue. La thèse a tout d'abord étudié le problème de stabilisation du drone sous l'hypothèse de petits déplacements (hypothèse de linéarité). Dans un second temps, on a montré comment relâcher l'hypothèse de petits déplacements via la synthèse de commandes non linéaires. Le cas du suivi de trajectoire a ensuite été considéré, en s'appuyant sur la définition d'un cadre générique de mesure d'erreur de position par rapport à un point de référence inconnu. Enfin, la validation expérimentale de ces résultats a été entamée pendant la thèse, et a permis de valider bon nombre d'étapes et de défis associés à leur mise en œuvre en conditions réelles. La thèse se conclut par des perspectives pour poursuivre les travaux. / The computers miniaturization has paved the way for the conception of Unmanned Aerial vehicles - "UAVs"- that is: flying vehicles embedding computers to make them partially or fully automated for such missions as e.g. cluttered environments exploration or replacement of humanly piloted vehicles for hazardous or painful missions. A key challenge for the design of such vehicles is that of the information they need to find in order to move, and, thus, the sensors to be used in order to get such information. A number of such sensors have flaws (e.g. the risk of being jammed). In this context, the use of a videocamera offers interesting prospectives. The goal of this PhD work was to study the use of such a videocamera in a minimal sensors setting: essentially the use of visual and inertial data. The work has been focused on the development of control laws offering the closed loop system stability and robustness properties. In particular, one of the major difficulties we faced came from the limited knowledge of the UAV environment. First we have studied this question under a small displacements assumption (linearity assumption). A control law has been defined, which took performance criteria into account. Second, we have showed how the small displacements assumption could be given up through nonlinear control design. The case of a trajectory following has then been considered, with the use of a generic error vector modelling with respect to an unknown reference point. Finally, an experimental validation of this work has been started and helped validate a number of steps and challenges associated to real conditions experiments. The work was concluded with prospectives for future work.

During this master thesis an adaptive control system for angular velocity was developed. The approach focuses on radio controlled quadrotors and thereby targets a computational efficient solution. The adaptation is divided into two steps. First, an Integral Squared Plus Time Delay (I2PD) model estimation is done through the Recursive Least Squares (RLS) algorithm. Second, control parameters for the quadrotors Proportional Integral Derivative (PID) controller are found by using empirically modified Tuning Rules (TRs) which are based on the I2PD model. Simulation results show the benefits of an adaptive tuning algorithm in regards of the time consuming process of manually tuning a PID controller as well as robustness against both predictable and unpredictable system changes. This includes actuator degradation, battery consumption, changes in weight or inertia etc. However, it is apparent that the tuning algorithm demands an arbitrary initial approximation to be stable at initialization.

Neste trabalho foi realizada a coleta de imagens de um trecho da área seca da UHE Álvaro de Souza Lima (Bariri – SP) e do lago da UNISINOS (São Leopoldo – RS), com câmeras e sensores, embarcados em RPAS (sistemas aéreos remotamente pilotados). As imagens resultantes foram processadas através da técnica estrutura do movimento (SfM), gerando ortomosaicos e modelos digitais de terrenos (MDTs). Pontos de checagem observados nestas imagens foram comparados com seus homólogos, obtidos por meio de posicionamento global diferencial no modo RTK. Testes estatísticos para verificação de hipóteses prévias de independência, aleatoriedade e normalidade, bem como, cálculos estatísticos e testes de inferência estatística foram realizados nas discrepâncias horizontais e verticais de ambos os produtos cartográficos. A verificação da exatidão vertical foi realizada somente nos produtos MDTs, a partir de feições de controle. O objetivo principal foi avaliar produtos cartográficos digitais gerados a partir de imagens coletadas por RPAS, quanto aos requisitos mínimos de exatidão geométrica para atendimento às leis estabelecidas pelos órgãos reguladores de energia elétrica. As discrepâncias horizontais observadas no ortomosaico da Unidade Hidroelétrica de Bariri apresentaram independência, não aleatoriedade e distribuição normal. Já para as discrepâncias verticais, foram independentes, aleatórias e normais. O teste t Student revelou tendência na posição horizontal e não tendência na vertical. O desvio padrão e o RMSE horizontais foram confrontados com o Padrão de Exatidão Cartográfica brasileiro através do teste Qui quadrado, indicando que os produtos têm precisão horizontal na escala 1:2000 (classe C) e, o MDT, na escala 1:10.000 (classe C), ambos com 95% de confiança e 22 e 19 graus de liberdade, respectivamente. A comparação entre MDTs conduziu a resultados semelhantes, porém demostrou que maiores discrepâncias ocorreram devido a fraca sobreposição de imagens. Para os produtos do lago UNISINOS, as discrepâncias cumpriram as hipóteses de independência e aleatoriedade. Porém, a normalidade não foi constatada pelo teste de Shapiro-Wilk. No entanto, através do teste U de Mann-Whitney, os produtos obtidos via sensor apresentaram semelhanças entre os pontos amostrados e seus homólogos. Através do teste direcional, os ângulos de direção foram medidos e caracterizaram a tendência horizontal para os ortomosaicos da UHE e do lago UNISINOS. Os resultados demonstraram que o mapeamento através de imagens via sensores embarcados em RPAS e processadas por SfM apresentaram exatidão posicional para integração com cartas de outras fontes, nas mesmas escalas. Portanto, estes produtos possibilitam a atualização de dados cartográficos para atender as demandas legais imputadas aos operadores de reservatórios. / In this work, the image collection was carried out on a stretch of dry area of the Alvaro de Souza Lima UHE (Bariri - SP) and of the UNISINOS lake (São Leopoldo) by cameras and sensors embedded in RPAS (Remotely Piloted Aerial System). The resulting images were processed using the Struture from Motion (SfM) technique in order to generate ortomosaicos and Digital Terrain Models (MDTs). Checkpoints observed in these images were compared with their counterparts obtained through differential global positioning in the RTK mode. Statistical tests for checking previous hypotheses of independence, randomness and normality, as well as statistical calculations and inference tests were conducted in the EN and h discrepancies of both cartographic products. The vertical accuracy verification was performed only in the MDTs products, from control features. The main objective of this work was to evaluate of digital cartographic products generated from photogrammetry collected by RPAS, with regard to the minimum requirements of geometric accuracy for fulfillment of the laws established by electric power regulatory agencies. The discrepancies observed in the orthomosaic of the Bariri Hydropower Plant exhibited independence, non randomness and normal distribution. However, for the MDT, these were independent, random, and normal. Students t-test showed a trend in the horizontal position and a non-trend in the vertical position. The horizontal standard deviation and RMSE (root mean square error) were analyzed according to the PEC-PCD Standard by the Chi-squared test, indicating that the orthomosaic fits at the 1:2,000 scale (class C), and the MDT, fits at the 1:10,000 scale (class C), both with a 95% confidence level and with degrees of freedom of 22 and 19, respectively. The comparison between MDTs leads to similar results, but showed that the largest discrepancies occur due to the weak overlap of the images. For the UNISINOS lake products, the discrepancies fulfilled the previous hypotheses of independence and randomness. But, the normality was not verified by the Shapiro-Wilk test. However, using the Mann-Whitney U test, the cartographic products obtained from a sensor showed similarities between the sampling points and their counterparts. By the directional test, it was measured the azimuth angles which characterizes the horizontal trend for the UHE and UNISINOS lake orthomosaics. The results demonstrated that the mapping by using images collected from sensors embedded in RPAS and processed through SfM presents positional accuracy for integration with maps from other sources, on the same scales. And therefore allows the updating of cartographic data map to meet the legal demands attributed to the reservoir operators.