UAV:er används frekvent i samhället och med detta har den kommersiella marknaden växt. Därför är det rimligt att de används i större utsträckning i konflikter, vilket innebär att konflikter där båda parter har UAV:er som kan klassas som relativt högteknologiska möts blir troligare. Ukraina och Rysslands användande av UAV i Ukraina kan räknas som denna typ av konflikt.  I uppsatsen är det UAV:er av den militära typen som diskuteras. Skillnaden mellan dessa och civila typer är framförallt räckvidd, flygtid och kvalitén på sensorerna.För att undersöka hur UAV:er nyttjas och taktiseras med i denna typ av konflikter har följande frågeställning använts: Hur påverkas nyttjandet av UAV:er i en konflikt mellan två högteknologiska motståndare?Genom att analysera beslutsprocessen med hjälp av OODA-loopen och bekämpningskedjan har författaren kunnat dra följande slutsatser om nyttjandet i denna typ av konflikt. Uppsatsen är genomförd som en fallstudie där metoderna kvalitativ textanalys och intervju använts Slutsatsen är att den multiplikator som UAV varit i Ukraina visar på att de kommer fortsätta användas i framtida konflikter. Trots att telekrig varit aktivt mot just UAV:erna och att de saknar motmedel mot störningen har de fortsatt att nyttjas, den multiplikatoreffekt de bidrar med kan motiveras stridsekonomiskt och väger tyngre än de problem som störningen innebär. / UAV: s are in more frequent use as a result of a growing commercial market. This increases the probability of UAV: s in conflicts. This means that conflicts where both sides have access to UAV: s that are relatively high-tech becomes more likely. Ukraine and Russia’s use of UAV: s in Ukraine can be described as this kind of conflict. In this paper it is primarily military UAV: s that are discussed. The difference between these and their civilian counterparts are range, flight time and the quality of the sensors. In order to understand how the use of UAV: s is being affected, the following question needs to be answered.  How is the use of UAV: s affected in a conflict between two high-tech opponents? This was answered by analysing the decision-making process using the OODA loop and the kill chain. The paper is a case study which uses qualitative text analysis and an interview.The conclusion of this paper is that UAV: s has acted as a force multiplier in Ukraine and they will be used in future conflicts. Despite the electronic warfare against the UAV: s and the fact that they are missing systems for counteracting the disturbance both sides continue to use UAV: s. The force multiplier that is gained from using UAV: s is justified from a battle economic standpoint despite being hindered by electronic warfare.

A forced landing for an Unmanned Aerial Vehicle (UAV) is required if there is an emergency on board that requires the aircraft to land immediately. Piloted aircraft in the same scenario have a human on board that is able to engage in the complex decision making process involved in the choice of a suitable landing location. If UAVs are to ever fly routinely in civilian airspace, then it is argued that the problem of finding a safe landing location for a forced landing is an important unresolved problem that must be addressed. This thesis presents the results of an investigation into the feasibility of using machine vision techniques to locate candidate landing sites for an autonomous UAV forced landing. The approach taken involves the segmentation of the image into areas that are large enough and free of obstacles; classification of the surface types of these areas; incorporating slope information from readily available digital terrain databases; and finally fusing these maps together using a high level set of simple linguistic fuzzy rules to create a final candidate landing site map. All techniques were evaluated on actual flight data collected from a Cessna 172 flying in South East Queensland. It was shown that the use of existing segmentation approaches from the literature did not provide the outputs required for this problem in the airborne images encountered in the gathered dataset. A simple method was then developed and tested that provided suitably sized landing areas that were free of obstacles and large enough to land. The advantage of this novel approach was that these areas could be extracted from the image directly without solving the difficult task of segmenting the entire image into the individual homogenous objects. A number of neural network classification approaches were tested with the surface types of candidate landing site regions extracted from the aerial images. A number of novel techniques were developed through experimentation with the classifiers that greatly improved upon the classification accuracy of the standard approaches considered. These novel techniques included: automatic generation of suitable output subclasses based on generic output classes of the classifier; an optimisation process for generating the best set of input features for the classifier based on an automated analysis of the feature space; the use of a multi-stage classification approach; and the generation of confidence measures based on the outputs of the neural network classifiers. The final classification result of the system performs significantly better than a human test pilot's classification interpretation of the dataset samples. In summary, the algorithms were able to locate candidate landing site areas that were free of obstacles 92.3 ±2.6% (99% confidence in the result) of the time, with free obstacle candidate landing site areas that were large enough to land in missed only 5.3 ±2.2% (99% confidence in the result) of the time. The neural network classification networks developed were able to classify the surface type of the candidate landing site areas to an accuracy of 93.9 ±3.7% (99% confidence in the result) for areas labelled as Very Certain. The overall surface type classification accuracy for the system (includes all candidate landing sites) was 91.95 ±4.2% (99% confidence in the result). These results were considered to be an excellent result as a human test pilot subject was only able to classify the same data set to an accuracy of 77.24 %. The thesis concludes that the techniques developed showed considerable promise and could be used immediately to enhance the safety of UAV operations. Recommendations include the testing of algorithms over a wider range of datasets and improvements to the surface type classification approach that incorporates contextual information in the image to further improve the classification accuracy.

<p dir="ltr">In this work, first, unsupervised machine learning is proposed as a study for detecting and classifying jamming attacks targeting unmanned aerial vehicles (UAV) operating at a 2.4 GHz band. Three scenarios are developed with a dataset of samples extracted from meticulous experimental routines using various unsupervised learning algorithms, namely K-means, density-based spatial clustering of applications with noise (DBSCAN), agglomerative clustering (AGG) and Gaussian mixture model (GMM). These routines characterize attack scenarios entailing barrage (BA), single- tone (ST), successive-pulse (SP), and protocol-aware (PA) jamming in three different settings. In the first setting, all extracted features from the original dataset are used (i.e., nine in total). In the second setting, Spearman correlation is implemented to reduce the number of these features. In the third setting, principal component analysis (PCA) is utilized to reduce the dimensionality of the dataset to minimize complexity. The metrics used to compare the algorithms are homogeneity, completeness, v-measure, adjusted mutual information (AMI) and adjusted rank index (ARI). The optimum model scored 1.00, 0.949, 0.791, 0.722, and 0.791, respectively, allowing the detection and classification of these four jamming types with an acceptable degree of confidence.</p><p dir="ltr">Second, following a different study, supervised learning (i.e., random forest modeling) is developed to achieve a binary classification to ensure accurate clustering of samples into two distinct classes: clean and jamming. Following this supervised-based classification, two-class and three-class unsupervised learning is implemented considering three of the four jamming types: BA, ST, and SP. In this initial step, the four aforementioned algorithms are used. This newly developed study is intended to facilitate the visualization of the performance of each algorithm, for example, AGG performs a homogeneity of 1.0, a completeness of 0.950, a V-measure of 0.713, an ARI of 0.557 and an AMI of 0.713, and GMM generates 1, 0.771, 0.645, 0.536 and 0.644, respectively. Lastly, to improve the classification of this study, semi-supervised learning is adopted instead of unsupervised learning considering the same algorithms and dataset. In this case, GMM achieves results of 1, 0.688, 0.688, 0.786 and 0.688 whereas DBSCAN achieves 0, 0.036, 0.028, 0.018, 0.028 for homogeneity, completeness, V-measure, ARI and AMI respectively. Overall, this unsupervised learning is approached as a method for jamming classification, addressing the challenge of identifying newly introduced samples.</p>

With the expected growth of Unmanned Aerial Vehicle (UAV) traffic in the coming years, the demand for UAV tracking solutions in the Air Traffic Control (ATC) industry has been incentivized. To ensure the safe integration of UAVs into airspace, Air Traffic Management (ATM) systems will need to provide a number of services such as UAV tracking. The Interacting Multiple Model Extended Kalman Filter (IMM-EKF) is an industry standard for aircraft tracking, but no such algorithm has been tried and tested for UAV tracking. This thesis aims to determine a suitable tracking algorithm for the specific case of non-cooperative tracking of UAVs from radar data. In non-cooperative tracking scenarios, we do not have any information regarding the UAV other than radar measurements indicating the target’s position. We investigate an Adaptive Interacting Multiple Model Unscented Kalman Filter (IMM-UKF) method with three different motion model combinations in addition to comparing a Cartesian vs. Spherical measurement model. A comparison of motion models shows that using a Constant Jerk (CJ) model to model target maneuvers in the IMM structure reduces the risk of filter divergence as compared to using a turn model, such as Constant Turn (CT) or Constant Angular Velocity (CAV). The CJ model is thus a suitable choice to have as one of the motion models in an IMM structure and works well in conjunction with two Constant Velocity (CV) models. We were not able to determine if the Spherical measurement model is better than the Cartesian measurement model in general. However, the Spherical measurement model improves the accuracy of the state estimate in some cases. Adaptive tuning of the system noise covariance Q and measurement noise covariance R does not improve the accuracy of the state estimate but it improves the filter robustness and consistency when the filter is incorrectly tuned. Based on our results, we believe that the adaptive IMM-UKF shows promise but that there is still room for improvement with regards to both the accuracy and consistency. However, we will need to perform extensive tests with real UAV radar data to draw concrete conclusions. / Med den förväntade tillväxten av trafik med obemannade flygfordon (UAV) under de kommande åren kommer efterfrågan för spårningslösningar för UAV inom flygövervakning. För att säkerställa en säker integration av UAV:er i luftrummet, kommer Air Traffic Management (ATM)-system att behöva tillhandahålla tjänster för UAV-spårning. Det så kallade Interacting Multiple Model Extended Kalman Filter (IMM-EKF) filtret är en industristandard spårning av flygplan, men ingen sådan algoritm har prövats och testats för UAV-spårning. Denna avhandling syftar till att fastställa en lämplig spårningsalgoritm för det specifika fallet med icke samarbetande spårning av UAV från radardata. I icke samarbetande spårningsscenarier har vi ingen information om UAV:n utöver radarmätningar. Vi presenterar en adaptiv metod baserad på IMM-UKF, där vi ersätter EKF i industristandarden IMM-EKF med ett filter av typen UKF. Vi undersöker tre olika kombinationer av rörelsemodeller och jämför också en kartesisk med en sfärisk mätmodell. Vår jämförelse av rörelsemodeller visar om man använder en Constant Jerk (CJ) modell för manövrar i IMM-strukturen minskar risken för divergens jämfört med att använda en svängmodell, såsom Constant Turn (CT) eller Constant Angular Velocity (CAV). CJ-modellen är alltså ett lämpligt val att ha som en av rörelsemodellerna i en IMM-struktur och fungerar bra i kombination med två Constant Velocity (CV) modeller. Vi kunde inte avgöra om den sfäriska modellen var bättre än den kartesiska modellen. Adaptiv inställning av systembrusets kovarians Q och mätbrus kovarians R förbättrar inte tillståndsuppskattningens noggrannhet men den förbättrar filtrets robusthet och konsistens när filtret är felaktigt inställt. Baserat på våra resultat tror vi att den adaptiva IMM-UKF metoden är lovande men att det fortfarande finns utrymme för förbättringar när det gäller både noggrannhet och konsistens i spårningen. Vi kommer dock att behöva utföra omfattande tester med riktiga UAV-radardata för att dra konkreta slutsatser.

<p dir="ltr">This Research focuses on developing artificial intelligence tools to detect and mitigate cyber-attacks targeting unmanned aerial vehicles. </p>

The miniature tailsitter is a unique aircraft with inherent advantages over typical unmanned aerial vehicles. With the capabilities of both hover and level flight, these small, portable systems can produce efficient maneuvers for enhanced surveillance and autonomy with little threat to surroundings and the system itself. Such vehicles are accompanied with control challenges due to the two different flight regimes. Problems with the conventional attitude representation arise in estimation and control as the system departs from level flight conditions. Furthermore, changing dynamics and limitations in modeling and sensing give rise to significant attitude control design challenges. Restrictions in computation also result from the limited size and weight capacity of the miniature airframe. In this research, the inherent control challenges discussed above are addressed with a computationally efficient adaptive quaternion control algorithm. A backstepping method for model cancellation and consistent tracking of reference model attitude dynamics is derived. This is used in conjunction with two different algorithms designed for the identification of system parameters. For a metric of baseline performance, gain-scheduled quaternion feedback control is developed. With a regularized data-weighting recursive least-squares parameter estimation algorithm, the adaptive quaternion controller is shown to be better than the baseline method in simulation and hardware results. This method is also shown to produce universal performance for all aircraft with the three conventional control surface actuators (aileron, elevator, and rudder) barring saturation and assuming accurate system identification. Testing of attitude control algorithms requires development in quaternion-based navigational control and attitude estimation. A novel technique for hover north/east position control is derived. Also, altitude tracking in hover, given an inconsistent thrust system, is addressed with an original method of on-line throttle system identification. Means for quaternion-based level flight control are produced from adaptations made to existing techniques employed in the Brigham Young University Multi-Agent Coordination and Control Lab. Also generated are simple trajectories for transitions between flight modes. A method for the estimation of quaternion attitude is developed, which uses multiple sensors combined in a filtering technique similar to the fixed-gain Kalman filter. Simulation and hardware results of these methods are presented for concept validation. A discussion of the development and production of these testing means (a simulation environment and hardware flight test system) is provided. In culmination, a fully autonomous miniature tailsitter system is produced with results demonstrating its various capabilities.