Two micro air vehicles, namely Sapthami and Sapthami-flyer, are developed in this thesis. Their total weight is less than 200grams each. They fit inside a 30cm and 32cm sphere respectively and carry the commercially available Kestrel autopilot hardware. The vehicles have an endurance of around 20-30 minutes. The stall speed of Sapthami is around 7m/s and that of Sapthami-flyer is around 5m/s as found by nonlinear modeling. The low stall speed makes it possible for them to be launched by hand. This enhances their portability as they do not require any launching equipment. The vehicle installed with Kestrel autopilot system is capable of many modes of operations. It is capable of fully autonomous flight with the aid of a variety of sensors like the GPS unit, heading sensor, 2-axis magnetometer, 3-axis accelerometer and 3-axis gyros. The vehicle carrying the Kestrel autopilot hardware is capable of autonomous and semi-autonomous flights after installation and tuning of feedback loops. Sapthami, is a tailless flying wing with an inverse zimmermann profile. A flying wing is a preferred configuration for the MAV as it maximizes the lifting area for a given size constraint. For a maximum size constraint of 30cm and aspect ratio around 1, the vehicle operates at Reynolds number between 100,000 to 250,000, at flight velocity 7 m/s to 15 m/s. The Inverse Zimmerman profile has a higher lift coefficient, CL, in comparison to the other planforms such as rectangular, elliptical and Zimmermann, for aspect ratio 1 to 1.25 and tested at Reynolds number of 100,000. The configuration of Sapthami is clean as there is no fuselage and all the components like autopilot hardware and battery are housed inside the wing. A thick reflex Martin Hepperle (MH) airfoil MH18 is chosen which gives sufficient space to place the components. This airfoil is specially used for tailless configurations due to its negative camber at the trailing edge. This negative camber helps in reducing the negative pitching moment of the wing, since no separate horizontal tail is available on a tailless aircraft to compensate for it. The vehicle is fabricated using the blue foam, having a density of 30kg/m3 . The wing is fabricated by CNC machining after which slots are cut manually to embed the electronics. The vehicle is found to have stable flying characteristics. Limited flight trials are done for Sapthami. It takes large time to fabricate the vehicle due to limited availability of CNC machining facility. Therefore, a new tailless, wing-fuselage configuration, which can be fabricated with balsa wood, is designed. Sapthami-flyer is the second vehicle designed in this thesis. Its wing span is slightly more than Sapthami. Since it is a wing-fuselage configuration, therefore there is no need for a thick airfoil. Mark drela’s AG airfoils are found to have better lift than MH airfoils for the inverse Zimmerman profile. The thickness of the airfoils is reduced to 1% so that the wing can be made by a 1.5mm thick balsa sheet to reduce weight. The inverse Zimmermann profile wing with the AG09 airfoil is found to have best lift-to-drag ratio when compared to AG36, MH45 and MH18. The analysis is done using commercially available AVL software. AG09 with 1% thickness is used in the final configuration. This configuration has better short period damping than Sapthami. It also has slower modes of operation than Sapthami. The operating modes of most of the MAVs, including Sapthami and Sapthami-flyer, are lowly damped but fast. This makes it difficult for the pilot to fly the vehicle. To improve the flying qualities of the vehicle artificial stabilization is required. The feedback is implemented on the Kestrel autopilot hardware. It allows only PID based feedback structures to be implemented, hence gives no choice to the designer to implement higher order control. The digital integrator and differentiator implementation for feedback are non-ideal. This further reduces the effectiveness of control. The problem is dealt with by incorporating the additional dynamics introduced by these implementation while formulating the control problem. Further the modeling of the micro air vehicle is done by using vortex lattice simulation based softwares. The fidelity of the obtained dynamics is low. Therefore, there is high uncertainty in the plant model. The controller also needs to reject the wind gust disturbances which are of the order of the flight speed of the vehicle. All the above stated requirements from the control design can be best addressed by a robust control design. Sapthami-flyer uses aileron and elevator for control. There is no rudder in the configuration in order to reduce weight. In the longitudinal dynamics, pitch rate and pitch error feedback to elevator are used to increase the short period and phugoid damping respectively. In the lateral dynamics, a combination of roll rate, yaw rate and roll error feedback is given to aileron to improve the dutch roll damping and stabilize the spiral mode. The feedback loops for both longitudinal and lateral dynamics are multi-output single input design problems, therefore simultaneous tuning of loops is beneficial. The PID control is designed by first converting the actual plant to a static output feedback equivalent plant. The dynamics introduced by non-ideal differentiator and integrator implementation on the autopilot hardware are incorporated in the open loop static output feedback formulation. The robust pole placement for the SOF plant is done by using the modified iterative matrix inequality algorithm developed in this thesis. It is capable of multi-loop, multi-objective feedback design for SOF plants. The algorithm finds the optimal solution by simultaneously putting constraints on the H2 performance, pole placement, gain and phase margin of the closed loop system. The pole placement is done to minimize the real part of largest eigenvalue. A single controller is designed at a suit-able operating point and constraints are put on the gain and phase margin of the closed loop plant at other operating points. The designed controller is tested in flight on board Sapthami-flyer. The vehicle is also capable of tracking commanded pitch and roll attitude with the help of pitch error, roller or feedbacks. This is shown in the flight when the pilot leaves the RC stick and the vehicle tracks the desired attitude. The vehicle has shown improved flying characteristics in the closed loop mode.

Parameterised analytical models that describe the trimmed inflight behaviour of classical aircraft have been studied and are widely accepted by the flight dynamics community. Therefore, the primary role of aircraft parameter estimation is to quantify the parameter values which make up the models and define the physical relationship of the air vehicle with respect to its local environment. Nevertheless, a priori empirical predictions dependent on aircraft design parameters also exist, and these provide a useful means of generating preliminary values predicting the aircraft behaviour at the design stage. However, at present the only feasible means that exist to actually prove and validate these parameter values remains to extract them through physical experimentation either in a wind-tunnel or from a flight test. With the advancement of UAVs, and in particular smaller UAVs (less than 1m span) the ability to fly the full scale vehicle and generate flight test data presents an exciting opportunity. Furthermore, UAV testing lends itself well to the ability to perform rapid prototyping with the use of COTS equipment. Real-time system identification was first used to monitor highly unstable aircraft behaviour in non-linear flight regimes, while expanding the operational flight envelope. Recent development has focused on creating self-healing control systems, such as adaptive re-configurable control laws to provide robustness against airframe damage, control surface failures or inflight icing. In the case of UAVs real-time identification, would facilitate rapid prototyping especially in low-cost projects with their constrained development time. In a small UAV scenario, flight trials could potentialy be focused towards dynamic model validation, with the prior verification step done using the simulation environment. Furthermore, the ability to check the estimated derivatives while the aircraft is flying would enable detection of poor data readings due to deficient excitation manoeuvres or atmospheric turbulence. Subsequently, appropriate action could then be taken while all the equipment and personnel are in place. This thesis describes the development of algorithms in order to perform online system identification for UAVs which require minimal analyst intervention. Issues pertinent to UAV applications were: the type of excitation manoeuvers needed and the necessary instrumentation required to record air-data. Throughout the research, algorithm development was undertaken using an in-house Simulink© model of the Aerosonde UAV which provided a rapid and flexible means of generating simulated data for analysis. In addition, the algorithms were further tested with real flight test data that was acquired from the Cranfield University Jestream-31 aircraft G-NFLA during its routine operation as a flying classroom. Two estimation methods were principally considered, the maximum likelihood and least squares estimators, with the aforementioned found to be best suited to the proposed requirements. In time-domain analysis reconstruction of the velocity state derivatives ˙W and ˙V needed for the SPPO and DR modes respectively, provided more statistically reliable parameter estimates without the need of a α- or β- vane. By formulating the least squares method in the frequency domain, data issues regarding the removal of bias and trim offsets could be more easily addressed while obtaining timely and reliable parameter estimates. Finally, the importance of using an appropriate input to excite the UAV dynamics allowing the vehicle to show its characteristics must be stressed.

<div>Good wind-tunnel flow quality characteristics are vital to using test data in the aerodynamic design process. Spatially uniform velocity profiles are required to avoid yaw and roll moments that would not be present in real flight conditions. Low turbulence intensity levels are also important as several aerodynamic properties are functions of turbulence intensity. When measuring mean flow and turbulence properties, hot-wire anemometry offers good spatial resolution and high-frequency response with a fairly simple operation, and the ability to make near-wall measurements. Using hot-wire anemometry, flow quality experiments were conducted</div><div>in a closed-circuit wind tunnel with a test section that has a cross section area of 1.2 m x 1.8 m (4 ft. x 6 ft.). The experiments included measurements of flow velocity and turbulence intensity variation over the test section cross-section, spatial and temporal temperature variation, and</div><div>boundary layer measurements. The centerline velocity and turbulence intensity were also measured for flow speeds ranging from 13 to 43 m/s.</div>

A área de dinâmica de aeronaves atingiu um alto nível de desenvolvimento e devido à crescente disponibilidade de computadores cada vez mais rápidos e com maior capacidade de processamento; a aplicação de técnicas numéricas de identificação nesta área também teve grande avanço. Este trabalho apresenta uma metodologia para predição de respostas de aeronaves dentro de envelopes de vôo pré-estabelecidos usando redes neurais recorrentes e uma metodologia para estimação das suas derivadas aerodinâmicas usando redes neurais feedforward. Para obter os conjuntos de dados para treinar as redes neurais, foi implementado um modelo não linear de dinâmica de vôo e simulado o comportamento de uma aeronave de combate em nove pontos de um envelope de vôo. Foram usadas as respostas simuladas correspondentes a quatro pontos para treinar a rede neural e depois disto, esta capturou satisfatoriamente a dinâmica da aeronave, identificando com grande sucesso as respostas do movimento longitudinal da aeronave por todo o envelope de vôo considerado. Após a simulação e identificação das respostas da aeronave dentro do envelope de vôo, é apresentada a resolução do problema inverso, ou seja, usando velocidades escalares e angulares da aeronave juntamente com seus dados geométricos como entradas para a rede neural feedforward, é obtido um modelo neural estimador de derivadas aerodinâmicas. Para mostrar a capacidade deste modelo neural estimador, este é aplicado na estimação das derivadas da aeronave simulada e também aplicado na estimação das derivadas aerodinâmicas da aeronave militar a jato Xavante AT-26 da Força Aérea Brasileira. Estas metodologias propostas reduzem custo de obtenção das derivadas aerodinâmicas e mostram a eficácia das redes neurais em estimar as respostas de aeronaves dentre de um envelope de vôo pré-definido. / The area of aircraft dynamics has reached a high level of development and due to the increasing availability of computers continuously faster and with bigger processing capacity, the application of numerical identification techniques in this area also had great advance. This work presents two methodologies, one for prediction of aircraft responses within a pre-established flight envelope using recurrent neural networks and another one for estimation of its aerodynamic derivatives using feedforward neural networks. To get data sets to train the neural networks, a combat aircraft flight dynamics non-linear model was implemented and simulated in nine points of the flight envelope to obtain its behavior. The simulated responses corresponding to a four points of the flight envelope were used to train the neural network and after that, it was possible to verify that this net satisfactorily captured the dynamics of the aircraft, identifying with great success the longitudinal motion responses of the aircraft at all the considered flight envelope positions. After the simulation and identification of the aircraft responses inside the flight envelope, the solution of the inverse problem is presented, i.e., using scalar and angular aircraft velocities together with its geometric data as input to the feedforward neural network, a neural estimator model of aerodynamic derivatives is obtained. In order to show the capacity of this neural estimator model, this model is applied to the estimation of the derivatives of the simulated aircraft as well as to the estimation of the aerodynamic derivatives of a brazilian air force military jet aircraft, the Xavante AT-26. These proposed methodologies reduce the cost of obtaining the aerodynamic derivatives and show the estimation effectiveness of the neural networks to estimate the responses of an aircraft inside a pre-defined flight envelope.

<p>Instrument Control Sweden (ICS) is a software company that develops NATO STANAG 4586 compatible ground station software for control of unmanned systems such as unmanned aerial vehicles (UAVs). To perform testing and demonstration of the ground station software ICS needs a realistic UAV simulator that implements the STANAG 4586 protocol. This thesis studies what methods are best suited for the rapid development of such a simulator.</p><p>One goal with the project was to examine what existing flight simulator systems and flight dynamics models can be used to rapidly develop a UAV simulator. Another goal was to design and implement such a simulator. It is found that it’s possible to quickly develop a UAV simulator based on existing projects such as the flight simulator FlightGear, the simulation framework OpenEaagles and the flight dynamics model (FDM) JSBSim.</p><p>The design of the simulator is modular, object-oriented and features real-time design techniques. The main application is a simulation of a Vehicle Specific Module, which implements the STANAG 4586 protocol. Another module based on the OpenEaagles framework simulates the aircraft and its subsystems. A third module consists of the JSBSim FDM and simulates the flight dynamics and movements of the aircraft under the forces and moments affecting it.</p>

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

Parameterised analytical models that describe the trimmed inflight behaviour of classical aircraft have been studied and are widely accepted by the flight dynamics community. Therefore, the primary role of aircraft parameter estimation is to quantify the parameter values which make up the models and define the physical relationship of the air vehicle with respect to its local environment. Nevertheless, a priori empirical predictions dependent on aircraft design parameters also exist, and these provide a useful means of generating preliminary values predicting the aircraft behaviour at the design stage. However, at present the only feasible means that exist to actually prove and validate these parameter values remains to extract them through physical experimentation either in a wind-tunnel or from a flight test. With the advancement of UAVs, and in particular smaller UAVs (less than 1m span) the ability to fly the full scale vehicle and generate flight test data presents an exciting opportunity. Furthermore, UAV testing lends itself well to the ability to perform rapid prototyping with the use of COTS equipment. Real-time system identification was first used to monitor highly unstable aircraft behaviour in non-linear flight regimes, while expanding the operational flight envelope. Recent development has focused on creating self-healing control systems, such as adaptive re-configurable control laws to provide robustness against airframe damage, control surface failures or inflight icing. In the case of UAVs real-time identification, would facilitate rapid prototyping especially in low-cost projects with their constrained development time. In a small UAV scenario, flight trials could potentialy be focused towards dynamic model validation, with the prior verification step done using the simulation environment. Furthermore, the ability to check the estimated derivatives while the aircraft is flying would enable detection of poor data readings due to deficient excitation manoeuvres or atmospheric turbulence. Subsequently, appropriate action could then be taken while all the equipment and personnel are in place. This thesis describes the development of algorithms in order to perform online system identification for UAVs which require minimal analyst intervention. Issues pertinent to UAV applications were: the type of excitation manoeuvers needed and the necessary instrumentation required to record air-data. Throughout the research, algorithm development was undertaken using an in-house Simulink© model of the Aerosonde UAV which provided a rapid and flexible means of generating simulated data for analysis. In addition, the algorithms were further tested with real flight test data that was acquired from the Cranfield University Jestream-31 aircraft G-NFLA during its routine operation as a flying classroom. Two estimation methods were principally considered, the maximum likelihood and least squares estimators, with the aforementioned found to be best suited to the proposed requirements. In time-domain analysis reconstruction of the velocity state derivatives ˙W and ˙V needed for the SPPO and DR modes respectively, provided more statistically reliable parameter estimates without the need of a α- or β- vane. By formulating the least squares method in the frequency domain, data issues regarding the removal of bias and trim offsets could be more easily addressed while obtaining timely and reliable parameter estimates. Finally, the importance of using an appropriate input to excite the UAV dynamics allowing the vehicle to show its characteristics must be stressed.

The field of mobile robotic systems has become a rich area of research and design. These systems can navigate difficult terrain using multiple actuators with conventional ambulation, by hopping, jumping, or for aerial vehicles, using flapping wings, propellers, or engines to maintain aerial flight. Unmanned Aerial Systems(UAS) have been used extensively in both military and civilian applications such as reconnaissance or search and rescue missions. For air vehicles, range and endurance is a crucial design parameter as it governs which missions can be performed by a particular vehicle. In addition, when considering the presence of external disturbances such as atmospheric winds, these missions can be even more challenging. Meta aircraft technologies is one area of research that can increase range and endurance by taking advantage of an increase in L/D. A meta aircraft is an aircraft composed of smaller individual aircraft connected together through a similar connection mechanism that can potentially transfer power, loads, or information. This dissertation examines meta aircraft flight dynamics and controls for a variety of different configurations. First, the dynamics of meta aircraft systems are explored with a focus on the changes in fundamental aircraft modes and flexible modes of the system. Specifically, when aircraft are connected, the fundamental modes change, can become overdamped or even unstable. In addition, connected aircraft exhibit complex flexible modes and mode shapes that change based on the parameters of the connection joint and the number of connected aircraft. Second, the connection dynamics are explored for meta aircraft where the vehicles are connected wing tip to wing tip using passive magnets with a particular focus on modeling the connection event between aircraft in a practical environment. It is found that a multi-stage connection control law with position and velocity feedback from GPS and connection point image feedback from a camera yields adequate connection performance in the presence of realistic sensor errors and atmospheric winds. Furthermore, atmosphericwinds with low frequency gusts at the intensity normally found in a realistic environment pose the most significant threat to the success of connection. The frequency content of the atmospheric disturbance is an important variable to determine success of connection. Finally, the geometry of magnets that create the connection force field can alter connection rates. Finally, the performance of a generic meta aircraft system are explored. Using a simplified rigid body model to approximate any meta aircraft configuration, adequate connection is achieved in the presence of realistic winds. Using this controller overall performance is studied. In winds, there is an overall decrease in outer loop performance for meta aircraft. However, inner loop performance increases for meta aircraft. In addition, the aerodynamic benefit of different configurations are investigated. Wing to wing tip connected flight provides the most benefit in terms of average increased Lift to Drag ratio while tip to tail configurations drop the Lift to Drag ratio as trailing aircraft fly in the downwash of the leading aircraft.

A área de dinâmica de aeronaves atingiu um alto nível de desenvolvimento e devido à crescente disponibilidade de computadores cada vez mais rápidos e com maior capacidade de processamento; a aplicação de técnicas numéricas de identificação nesta área também teve grande avanço. Este trabalho apresenta uma metodologia para predição de respostas de aeronaves dentro de envelopes de vôo pré-estabelecidos usando redes neurais recorrentes e uma metodologia para estimação das suas derivadas aerodinâmicas usando redes neurais feedforward. Para obter os conjuntos de dados para treinar as redes neurais, foi implementado um modelo não linear de dinâmica de vôo e simulado o comportamento de uma aeronave de combate em nove pontos de um envelope de vôo. Foram usadas as respostas simuladas correspondentes a quatro pontos para treinar a rede neural e depois disto, esta capturou satisfatoriamente a dinâmica da aeronave, identificando com grande sucesso as respostas do movimento longitudinal da aeronave por todo o envelope de vôo considerado. Após a simulação e identificação das respostas da aeronave dentro do envelope de vôo, é apresentada a resolução do problema inverso, ou seja, usando velocidades escalares e angulares da aeronave juntamente com seus dados geométricos como entradas para a rede neural feedforward, é obtido um modelo neural estimador de derivadas aerodinâmicas. Para mostrar a capacidade deste modelo neural estimador, este é aplicado na estimação das derivadas da aeronave simulada e também aplicado na estimação das derivadas aerodinâmicas da aeronave militar a jato Xavante AT-26 da Força Aérea Brasileira. Estas metodologias propostas reduzem custo de obtenção das derivadas aerodinâmicas e mostram a eficácia das redes neurais em estimar as respostas de aeronaves dentre de um envelope de vôo pré-definido. / The area of aircraft dynamics has reached a high level of development and due to the increasing availability of computers continuously faster and with bigger processing capacity, the application of numerical identification techniques in this area also had great advance. This work presents two methodologies, one for prediction of aircraft responses within a pre-established flight envelope using recurrent neural networks and another one for estimation of its aerodynamic derivatives using feedforward neural networks. To get data sets to train the neural networks, a combat aircraft flight dynamics non-linear model was implemented and simulated in nine points of the flight envelope to obtain its behavior. The simulated responses corresponding to a four points of the flight envelope were used to train the neural network and after that, it was possible to verify that this net satisfactorily captured the dynamics of the aircraft, identifying with great success the longitudinal motion responses of the aircraft at all the considered flight envelope positions. After the simulation and identification of the aircraft responses inside the flight envelope, the solution of the inverse problem is presented, i.e., using scalar and angular aircraft velocities together with its geometric data as input to the feedforward neural network, a neural estimator model of aerodynamic derivatives is obtained. In order to show the capacity of this neural estimator model, this model is applied to the estimation of the derivatives of the simulated aircraft as well as to the estimation of the aerodynamic derivatives of a brazilian air force military jet aircraft, the Xavante AT-26. These proposed methodologies reduce the cost of obtaining the aerodynamic derivatives and show the estimation effectiveness of the neural networks to estimate the responses of an aircraft inside a pre-defined flight envelope.

Instrument Control Sweden (ICS) is a software company that develops NATO STANAG 4586 compatible ground station software for control of unmanned systems such as unmanned aerial vehicles (UAVs). To perform testing and demonstration of the ground station software ICS needs a realistic UAV simulator that implements the STANAG 4586 protocol. This thesis studies what methods are best suited for the rapid development of such a simulator. One goal with the project was to examine what existing flight simulator systems and flight dynamics models can be used to rapidly develop a UAV simulator. Another goal was to design and implement such a simulator. It is found that it’s possible to quickly develop a UAV simulator based on existing projects such as the flight simulator FlightGear, the simulation framework OpenEaagles and the flight dynamics model (FDM) JSBSim. The design of the simulator is modular, object-oriented and features real-time design techniques. The main application is a simulation of a Vehicle Specific Module, which implements the STANAG 4586 protocol. Another module based on the OpenEaagles framework simulates the aircraft and its subsystems. A third module consists of the JSBSim FDM and simulates the flight dynamics and movements of the aircraft under the forces and moments affecting it.