Micro aerial vehicle development and atmospheric flight on Mars are areas that require research in very low Reynolds number flight. Facilities for studying these problems are not widely available. The upper atmosphere of the Earth, approximately 100,000 feet AGL, is readily available and closely resembles the atmosphere on Mars, in both temperature and density. This low density also allows normal size test geometry with a very low Reynolds number. This solves a problem in micro aerial vehicle development; it can be very difficult to manufacture instrumented test apparatus in the small sizes required for conventional testing. This thesis documents the design, construction, and testing of a glider designed to be released from a weather balloon at 100,000 feet AGL and operate in this environment, collecting airfoil and aircraft performance data. The challenges of designing a vehicle to operate in a low Reynolds number, low temperature environment are addressed.

Aircraft modeling and simulation have become increasingly important in the areas of pilot training, safety and aircraft design, especially for unmanned aerial vehicles (UAVs). A userriendly, easily expandable, nonlinear six degree-ofreedom aircraft simulator for the Xipiter X-2C Xawk UAV was created to address these issues. The simulator will allow pilots to have an opportunity to train and gain experience in flying the aircraft even before it leaves the ground. In addition, it will allow for design modifications or new aircraft designs to be evaluated before time and money are spent on their implementation. This work can also serve as the basis for the development of control systems for the aircraft, such as a control augmentation system or autopilot.

The motivation for this research is the need to improve performance of the autonomous flight of a tailsitter UAV. Tailsitter aircraft combine the hovering and vertical take-off and landing capability of a rotorcraft with the long endurance flight capability of a fixed-wing aircraft. The particular aircraft used in this research is the V-Bat, a tailsitter UAV with a conventional wing and the propeller and control surfaces located within a ducted-fan tail assembly. This research focuses on identifying the models and parameters of the V-Bat in hover and level flight as a basis for the design of the control systems for hover, level, and transition modes of flight.Models and parameters were identified from experimental data. Wind-tunnel tests, bench tests, and flight tests were performed in a variety of flight conditions. Wind tunnel tests yielded force and moment coefficients over the full flight envelope of the V-Bat. Models and parameters for longitudinal, lateral, and hover flight are presented. Bench tests were conducted to enhance understanding about the ducted-fan propulsion system and the effectiveness of the control surfaces. The thrust characteristics of the ducted fan were measured. Control derivatives were derived from force and moment measurements. Flight tests were completed to obtain dynamic models of the V-Bat in hover flight. Using frequency-domain system identification methods, frequency-response and transfer function models of roll, pitch, and yaw responses to aileron, elevator, and rudder control input were derived.The results obtained from these experimental tests were used to identify models and parameters of the V-Bat aircraft, giving insight into its behavior and enhancing the control analysis and simulation capabilities for this aircraft, thus providing the increased levels of understanding needed for autonomous flight.

No description available.

Unmanned aerial vehicles (UAVs) attract increasing attention for various wireless network applications by using UAVs' reliable line-of-sight (LoS) paths in air-ground connections and their flexible placement and movement. As such, the wireless network architecture is becoming three-dimensional (3D), incorporating terrestrial and aerial network nodes, which is more dynamic than the traditional terrestrial communications network. Despite the UAVs' advantages of high LoS path probability and flexible mobility, the challenges of UAV communications need to be considered in the design of integrated air-ground networks, such as spectrum sharing, air-ground interference management, energy-efficient and cost-effective UAV-assisted communications. On the other hand, in wireless networks, users request not only reliable communication services but also execute computation-intensive and latency-sensitive tasks. As one of the enabling technologies in wireless networks, edge computing is proposed to offload users' computation tasks to edge servers to reduce users' latency and energy consumption. However, this requires efficient utilization of both communication resources and computation resources. Furthermore, integrating UAVs into edge computing networks brings many benefits, such as enhancing offloading ability and extending offloading coverage region. This dissertation makes a series of fundamental contributions to UAVs and edge computing in wireless networks that include: 1) Reliable UAV communications, 2) Efficient edge computing schemes, and 3) Integration of UAV and edge computing. In the first contribution, we investigate UAV spectrum access and UAV swarm-enabled aerial reconfigurable intelligent surface (SARIS) for achieving reliable UAV communications. On the one hand, we study a 3D spectrum sharing between device-to-device (D2D) and UAVs communications. Specifically, UAVs perform spatial spectrum sensing to opportunistically access the licensed channels occupied by the D2D communications of ground users. The results show that UAVs' optimal spatial spectrum sensing radius can be obtained given specific network parameters. On the other hand, we study the beamforming and placement design for SARIS networks in downlink transmissions. We consider that the direct links between the ground base station (BS) and mobile users are blocked due to obstacles in the urban environment. SARIS assists the BS in reflecting the signals to randomly distributed mobile users. The results show that the proposed SARIS network significantly improves the weighted sum-rate for ground users, and the placement design plays an essential role in the overall system performance. In the second contribution, we develop a joint communication and computation resource allocation scheme for vehicular edge computing (VEC) systems. The full channel state information (CSI) in VEC systems is not always available at roadside units (RSUs). The channel varies fast due to vehicles' mobility, and it is pretty challenging to estimate CSI and feed back the RSUs for processing the VEC algorithms. To address the above problem, we introduce a large-scale CSI-based partial computation offloading scheme for VEC systems. Using deep learning and optimization tools, we minimize the users' energy consumption while guaranteeing their offloading latency and outage constraints. The results demonstrate that the introduced resource allocation scheme can significantly reduce the total energy consumption of users compared with other computation offloading schemes. In the third contribution, we present novel frameworks for integrating UAVs to edge computing networks to achieve improved computing performance. We study mobile edge computing (MEC) in air-ground integrated wireless networks, including ground computational access points (GCAPs), UAVs, and user equipment (UE), where UAVs and GCAPs cooperatively provide computation resources for UEs. The resource allocation algorithm is developed based on the block coordinate descent method by optimizing the subproblems of users' association, power control, bandwidth allocation, computation capacity allocation, and UAV placement. The results show the advantages of the introduced iterative algorithm regarding the reduced total energy consumption of UEs. Finally, we highlight directions for future works to advance the research presented in this dissertation and discuss its broader impact across the wireless networks industry and standard-making. / Doctor of Philosophy / The fifth-generation (5G) cellular network aims to achieve a high data rate by having greater bandwidth, deploying denser networks, and multiplying the antenna links' capacity. However, the current wireless cellular networks are fixed on the ground and thus pose many disadvantages. Moreover, the improved system performance comes at the cost of increased capital expenditures and operating expenses in wireless networks due to the enormous energy consumption at base stations (BS) and user equipment (UE). More spectrum and energy-efficient yet cost-effective technologies need to be developed in next-generation wireless networks, i.e., beyond-5G or sixth-generation (6G) networks. Recently, unmanned aerial vehicle (UAV) has attracted significant attention in wireless communications. Due to UAVs' agility and mobility, UAVs can be quickly deployed to support reliable communications, resorting to its line-of-sight-dominated connections in the air-ground channels. However, the sufficient available spectrum for extensive UAV communications is scarce, and the co-channel interference in air-air and air-ground connections need to be considered in the design of UAV networks. In addition to users' communication requests, users also need to execute intensive computation tasks with specific latency requirements. As such, edge computing has been proposed to integrate wireless communications and computing by offloading users' computation tasks to edge servers in proximity, reducing users' computation energy consumption and latency. Besides, integrating UAVs into edge computing networks makes efficient offloading schemes by leveraging the advantages of UAV communications. This dissertation makes several contributions that enhance UAV communications and edge computing systems performance, respectively, and present novel frameworks for UAV-assisted three-dimensional (3D) edge computing systems. This dissertation addresses the fundamental challenges in UAV communications, including spectrum sharing, interference management, UAV 3D placement, and beamforming, allowing broadband, wide-scale, cost-effective, and reliable wireless connectivity. Furthermore, this dissertation focuses on the energy-efficient vehicular edge computing systems and mobile edge computing systems, where the UAVs are applied to achieve 3D edge computing systems. To this end, various mathematical frameworks and efficient joint communication and computation resource allocation algorithms are proposed to design, analyze, optimize, and deploy UAV and edge computing systems. The results show that the proposed air-ground integrated networks can deliver spectrum-and-energy-efficient yet cost-effective wireless services, thus providing ubiquitous wireless connectivity and green computation offloading in the future beyond-5G or 6G wireless networks.

The task of railroad inspection is a tedious one. It requires a lot of skilled experts and long hours of frequent on-field inspection. Automated ground equipment systems that have been developed to address this problem have the drawback of blocking the rail service during inspection process. As an alternative, using aerial imagery from a UAV, Computer Vision and Machine Learning based techniques were developed in this thesis to analyze two kinds of defects on the rail tracks. The defects targeted were missing spikes on tie plates and cracks on ties. In order to perform this inspection, the rail region was identified in the image and then the tie plate and tie regions on the track were detected. These steps were performed using morphological operations, filtering and intensity analysis. Once the tie plate was localized, the regions of interest on the plate were used to train a machine learning model to recognize missing spikes. Classification using SVM resulted in an accuracy of around 96% and varied greatly based on the tie plate illumination and ROI alignment for Lampasas and Chickasha subdivision datasets. Also, many other different classifiers were used for training and testing and an ensemble method with majority vote scheme was also explored for classification. The second category of learning model used was a multi-layered neural network. The major drawback of this method was, it required a lot of images for training. However, it performed better than feature based classifiers with availability of larger training dataset. As a second kind of defect, tie conditions were analyzed. From the localized tie region, the tie cracks were detected using thresholding and morphological operations. A machine learning classifier was developed to predict the condition of a tie based on training examples of images with extracted features. The multi-class classification accuracy obtained was around 83% and there were no misclassifications seen between two extreme classes of tie condition on the test data. / Master of Science

We study the problem of Reinforcement Learning (RL) for Unmanned Aerial Vehicle (UAV) navigation with the smallest number of real world samples possible. This work is motivated by applications of learning autonomous navigation for aerial robots in structural inspec- tion. A naive RL implementation suffers from curse of dimensionality in large continuous state spaces. Gaussian Processes (GPs) exploit the spatial correlation to approximate state- action transition dynamics or value function in large state spaces. By incorporating GPs in naive Q-learning we achieve better performance in smaller number of samples. The evalua- tion is performed using simulations with an aerial robot. We also present a Multi-Fidelity Reinforcement Learning (MFRL) algorithm that leverages Gaussian Processes to learn the optimal policy in a real world environment leveraging samples gathered from a lower fidelity simulator. In MFRL, an agent uses multiple simulators of the real environment to perform actions. With multiple levels of fidelity in a simulator chain, the number of samples used in successively higher simulators can be reduced. / Master of Science / Increasing development in the field of infrastructure inspection using Unmanned Aerial Vehicles (UAVs) has been seen in the recent years. This thesis presents work related to UAV navigation using Reinforcement Learning (RL) with the smallest number of real world samples. A naive RL implementation suffers from the curse of dimensionality in large continuous state spaces. Gaussian Processes (GPs) exploit the spatial correlation to approximate state-action transition dynamics or value function in large state spaces. By incorporating GPs in naive Q-learning we achieve better performance in smaller number of samples. The evaluation is performed using simulations with an aerial robot. We also present a Multi-Fidelity Reinforcement Learning (MFRL) algorithm that leverages Gaussian Processes to learn the optimal policy in a real world environment leveraging samples gathered from a lower fidelity simulator. In MFRL, an agent uses multiple simulators of the real environment to perform actions. With multiple levels of fidelity in a simulator chain, the number of samples used in successively higher simulators can be reduced. By developing a bidirectional simulator chain, we try to provide a learning platform for the robots to safely learn required skills in the smallest possible number of real world samples possible.

Unmanned aerial vehicles (UAVs) are increasingly being used for both military and commercial applications to replace more costly and dangerous manned operations. Vehicles with vertical take-off and landing (VTOL) and hovering capabilities are of interest for functions such as surveillance and inspection where the ability to hold the position of the vehicle is desired. Ducted fan vehicles are of particular interest because of their high efficiency per unit diameter when compared to the more commonly seen multirotor vehicles. This makes ducted fan UAVs very well suited for size-constrained missions such as indoor inspection or urban reconnaissance. However, the advantages of ducted fans come at the cost of complex nonlinear dynamics which present challenging modeling and control problems. This thesis provides a detailed discussion of the instrumentation, modeling, and control of a ducted fan UAV. The dynamic model of the UAV is computed from a simplified parametric model. Unknown parameters of the model are found from system identification based on flight data. Synthesis of a linear state feedback controller based on this model is discussed, and it is demonstrated in hardware that this controller can effectively stabilize the vehicle. / Master of Science / Unmanned aerial vehicles (UAVs) are increasingly being used for both military and commercial applications to replace more costly and dangerous manned operations. Vehicles with vertical take-off and landing (VTOL) and hovering capabilities are of interest for functions such as surveillance and inspection where the ability to hold the position of the vehicle is desired. Ducted fan vehicles are of particular interest because of their high efficiency per unit diameter when compared to the more commonly seen multirotor vehicles. This makes ducted fan UAVs very well suited for size-constrained missions such as indoor inspection or urban reconnaissance. However, the advantages of ducted fans come at the cost of complex dynamics which present challenging modeling and control problems. This thesis provides a detailed discussion of the instrumentation, modeling, and control of a ducted fan UAV. The dynamic model of the UAV is computed from a simplified parametric model. Unknown parameters of the model are found from system identification based on flight data. Using this parametric model, development of a linear controller that uses feedback from the vehicle’s state is discussed, and it is demonstrated in hardware that this controller can effectively stabilize the vehicle.

This thesis investigates the possibility of using reinforcement learning (RL) techniques to create a flight controller for a quadrotor Micro Aerial Vehicle (MAV). A capable flight control system is a core requirement of any unmanned aerial vehicle. The challenging and diverse applications in which MAVs are destined to be used, mean that considerable time and effort need to be put into designing and commissioning suitable flight controllers. It is proposed that reinforcement learning, a subset of machine learning, could be used to address some of the practical difficulties. While much research has delved into RL in unmanned aerial vehicle applications, this work has tended to ignore low level motion control, or been concerned only in off-line learning regimes. This thesis addresses an area in which accessible information is scarce: the performance of RL when used for on-policy motion control. Trying out a candidate algorithm on a real MAV is a simple but expensive proposition. In place of such an approach, this research details the development of a suitable simulator environment, in which a prototype controller might be evaluated. Then inquiry then proposes a possible RL-based control system, utilising the Q-learning algorithm, with an adaptive RBF-network providing function approximation. The operation of this prototypical control system is then tested in detail, to determine both the absolute level of performance which can be expected, and the effect which tuning critical parameters of the algorithm has on the functioning of the controller. Performance is compared against a conventional PID controller to maximise the usability of the results by a wide audience. Testing considers behaviour in the presence of disturbances, and run-time changes in plant dynamics. Results show that given sufficient learning opportunity, a RL-based control system performs as well as a simple PID controller. However, unstable behaviour during learning is an issue for future analysis. Additionally, preliminary testing is performed to evaluate the feasibility of implementing RL algorithms in an embedded computing environment, as a general requirement for a MAV flight controller. Whilst the algorithm runs successfully in an embedded context, observation reveals further development would be necessary to reduce computation time to a level where a controller was able to update sufficiently quickly for a real-time motion control application. In summary, the study provides a critical assessment of the feasibility of using RL algorithms for motion control tasks, such as MAV flight control. Advantages which merit interest are exposed, though practical considerations suggest at this stage, that such a control system is not a realistic proposition. There is a discussion of avenues which may uncover possibilities to surmount these challenges. This investigation will prove useful for engineers interested in the opportunities which reinforcement learning techniques represent.

<p>Nya typer av krigföring, så som COIN-operationer (Counter insurgency), ställer nya krav på flygstridskrafterna. Utvecklingen av det bemannade stridsflyget har, om man jämför med UAV (Unmanned Aerial Vehicle), trots detta stått relativt stilla. Obemannat stridsflyg har samtidigt utvecklat förmågor som gör funktionen till en värdig konkurrent på det moderna slagfältet. Syftet med denna uppsats är att undersöka huruvida obemannat stridsflyg kan överta det bemannade stridsflygets plats inom ramen för COIN-operationer. Detta avser jag uppnå genom att undersöka hur väl de båda funktionerna svarar uppmot de krav som ställs på ett stridsflygplan i en operation med inslag av COIN. Utifrån detta kommer jag bedöma vilka uppdragstyper som kan samt eventuellt inte kan lösas av en obemannad respektive bemannad flygfarkost. Resultatet i uppsatsen visar att bemannat stridsflyg är bättre anpassat till att genomföra markmålsoperationer inom ramen för COIN-operationer. Främst på grund av en bättre omvärldsuppfattning. Vid uppdragstyper kopplade till flygunderrätteleinhämtning är dock det obemannade stridsflyget bättre anpassat. Främst på grund av en längre uthållighet. Vad det gäller kostnadsaspekten så visar analysen att brukskostnaderna för det obemannade stridsflyget är betydligt lägre än hos det bemannade stridsflyget. Den höga anskaffningskostnaden för ett nytt UAV-system talar dock till dess nackdel. Skyddet som är integrerat i de obemannade plattformarna är sämre än hos den bemannade konkurrenten. UAV får i mångt och mycket lita till sin storlek för skydd. Slutsatsen jag kan dra av arbetet är att bemannat stridsflyg fortfarande har en plats på slagfältet inom ramen för COIN-operationer. Om utvecklingen fortsätter i samma takt och system som förbättrar omvärldsuppfattning för operatören på UAV samt möjliggör ett bättre självskydd kopplat till plattformen kan resultatet snart vara ett annat.</p> / <p>New types of warfare, such as COIN (Counter Insurgency), operations, mean new requirements for the Air Force. The development of the manned combat aircraft has, in comparison with the UAV (Unmanned Aerial Vehicle), shown relativly slow progress. Unmanned combat aircraft has on the other hand developed capabilities that make it a worthy competitor on the modern battlefield. The purpose of this paper is to examine whether unmanned combat aircraft can outrival the manned combat aircraft, in COIN-operations. I will achieve this by studying how well the two functions are responding to the demands of a combat aircraft in an operation with elements of COIN. From this, I will assess the mission types that can and can not possibly besolved by an unmanned, or a manned combat aircraft. The results of this paper show that the manned fighter is better suited to carry out counter land operations within thescope of COIN-operations, mainly due to a better situational awareness. Due to longer endurance, the unmanned combat aircraft is better suited to carry out missions like reconnaissance and surveillance. In the cost aspect, the analysis shows that the cost of using the unmanned combat aircraft is much lower than that of the manned combat aircraft. The high purchase cost of a new UAV-system, however, speaks against it. The protection that is integrated into the unmanned platforms is worse than for its rival. UAVs must largely rely on their size for protection. The conclusion I can draw from this paper is that manned fighter aircraft still has a place on the battlefield as a part of COIN-operations. If the development continues at this pace and systems that improve the situational awareness of the operator of UAVs, and allows for better self protection linked to the platform, the result may soon be another.</p>