Climate change is increasingly influencing the aviation industry, motivating novel aircraft concepts such as the Flying V. However, unconventional configurations introduce new challenges in stability, control, and handling qualities. Additionally, aerodynamic model uncertainties and limited research on robust control for the Flying V highlight need for improved flight control systems. This study advances the conceptual maturity of the Flying V by developing a H∞ loop-shaping C* longitudinal controller designed to ensure robust stability and performance under aerodynamic uncertainty and input/output disturbances while targeting Level 1 handling qualities. The Flying V model is implemented in MATLAB/Simulink, before a one-degree-of-freedom (1DOF) controller is designed. This is extended to a two-degree-of-freedom (2DOF) architecture incorporating parametric uncertainty and multi-model synthesis to maintain stability across the flight envelope. The 2DOF controller improves tracking and maintains good disturbance rejection while largely meeting Level 1 handling quality requirements and maintaining robust stability over many test case simulations.

The urgent need to further reduce fuel burn and climate impact for next-generation aircraft drives wings to become lighter and with higher aspect ratios to reduce induced drag. The resulting increase in wing flexibility presents both an opportunity and numerous challenges, as such wings experience increased structural loads at the wing root and are more prone to aeroelastic instabilities such as flutter. Passive and active flutter suppression and load alleviation techniques therefore provide a promising solution to enable these wings without the subsequent weight increase. For active flutter suppression and load alleviation, it is particularly important to express the free-flying aeroservoelastic model as an interpretable state-space model with manageable order.

In this regard, the aircraft loads model in the DLR Loads Kernel is formally derived and refined to make it suitable for free-flying aeroservoelasticity, constituting several improvements over the original model. The model is strictly expressed as a closed-form, symbolic, monolithic state-space representation, rigorously derived in this work. Besides improved interpretability and extensibility, this also allows for easily porting the nonlinear state-space model from Python to MATLAB/Simulink, and therefore effectively bridges NASTRAN and Simulink, the industry standards for finite-element modeling and control design respectively. To further improve suitability for free-flying aeroservoelasticity, the spiral nature of the Sears function is approximated using cascaded gust zones using a Padé approximation, significantly reducing the number of disturbance inputs. Physical RFA is employed, with the resulting aerodynamic lag dynamics projected to lag force and moment dynamics, achieving a substantial reduction in lag states. Additionally, the model is augmented with actuator dynamics and an accelerometer sensor model.

As the state-space model requires a representative aircraft, the Embraer Benchmark Wing, used for research on aeroelastic tailoring, is systematically transformed into a free-flying finite-element model, enabling the generation of free-flying aeroservoelastic state-space models with different wing mass and stiffness distributions at various mass cases and in varying flight conditions. Using this aircraft and in the considered simulations, the derived model achieves a 100–400x improvement in simulation time and a 458x reduction in input file size, and thus eliminates the need to save model output. Gust inputs are reduced from n to 2, and lag states from n x p to (6 + m + s) x p, with n aerodynamic panels, p RFA poles, m flexible modes, and s monitoring forces and/or moments. At the same time, these refinements are shown to not compromise model behaviour, as results show excellent agreement. These improvements, in conjunction with the added actuator dynamics and sensor model, make the state-space model well-suited for future work in free-flying active flutter suppression, active load alleviation, and on the longer-term, control co-design.

Launch vehicles' reusability has been regaining a lot of interest in recent years. Therefore, to continue the development of these types of vehicles, it is important that they are as reliable as possible, even in the case of faults, to achieve their objective of consistent reusability.


Current literature does not offer a detailed analysis of the influence of different faults or an analysis of these allocation algorithms for reusable rockets in a crucial scenario of these rockets, such as the descent phase. Hence, this thesis focuses on the development and investigation of the advantages of a fault-tolerant control allocation solution for propulsion system faults, to be applied in the powered descent and landing phase of a clustered-engine reusable launch vehicle. Furthermore, the different types of faults and fault characteristics are analysed to understand their impact on a mission of this sort.

This thesis presents a hybrid control framework that combines Reinforcement Learning (RL) and Model Predictive Control (MPC) to achieve constraint-satisfying flutter suppression and load alleviation in flexible aircraft subject to turbulent gusts. During training, MPC computes safe input bounds using high-fidelity Linear Parameter Varying (LPV) models and long prediction horizons, exploiting known disturbances to accurately capture aeroelastic behavior. A Q-learning agent is trained to select control actions within these bounds, adapting to nonlinear dynamics and actuator delay. At deployment, the learned policy operates from a lightweight Q-table, with certified interpolation ensuring constraint satisfaction even for unseen states. By integrating the anticipatory capabilities of MPC with the adaptability of RL, the framework enables effective control under turbulence and structural uncertainty. While demonstrated on an aeroelastic aircraft, the approach can be generalized to other systems with similar dynamics, where constraint handling and adaptive control under uncertainty are critical.

Following an extensive evaluation of subsystem options outlined in the mid-term report, the multi-rotor swarm configuration was selected as the final design, accompanied by a preliminary operational strategy. This decision has guided the project into its final detailed design phase. This report demonstrates how the design proceeds using the selected analytical methods. A series of detailed analyses further illustrates how the design meets the updated performance requirements, supported by numerical results and visualisations. Finally, post-project actions and future plans are outlined to ensure the project’s continued development and success.

In nature, birds can intelligently adapt their wing shapes to their environment. This paper aims to replicate this capability by designing an online data-driven aerodynamic performance optimization framework for an unconventional morphing aircraft. Compared to state-of-the-art methods, the proposed framework efficiently finds global optima with reduced computational load when addressing time-varying, nonlinear, and non-convex problems. It also demonstrates enhanced adaptability to unforeseen scenarios. In the event of a sudden actuator fault, the algorithm can automatically detect the fault, adapt the onboard data-driven model, and continue performing optimization and trimming tasks using the remaining healthy actuators. Additionally, the paper addresses the optimal number of actuators within a morphing surface, considering the tradeoff between optimization performance and the weight penalty. High-fidelity simulations demonstrate that through active morphing, the proposed framework achieves drag reductions of 1.9–4.9 % during cruise and up to 12.6 % at higher operational lift coefficients (due to heavier weight and lower speed), resulting in an overall drag reduction of 2.98 % over a typical flight cycle, which corresponds to fuel savings of approximately 150 kg/h. This research represents a significant advancement in sustainable aviation, contributing to reduced fuel consumption, lower emissions, and improved fault tolerance for next-generation aircraft.

Underactuated mechanical systems (UMS) feature prominently in robotics and aerospace, with aircraft, unmanned air vehicles, and aeroelastic wings as prime examples. These systems present multifaceted control challenges, ranging from inherent underactuation and stability concerns to state and control saturation and an overarching need for robustness. Crucially, reducing model dependency is a key strategy to enhance control robustness. Nonlinear Model Predictive Control (NMPC) is valued for addressing underactuation and constraints within complex nonlinear dynamics while considering future stages for immediate decision-making. However, implementing NMPC in UMS can pose challenges due to model uncertainties and external disturbances. To enhance NMPC's robustness in UMS control, we introduce a disturbance rejection NMPC strategy, which is tightly coupled with the incremental nonlinear dynamic inversion (INDI), a sensor-based adaptive control approach. The INDI is expected to reject most of the disturbances. Any disturbance residues are managed within a robust NMPC framework through constraint tightening. The efficacy of our method is exemplified through its application to two distinct UMS models. The proposed controller is first customized for a nonlinear aeroelastic system. Compared to the nominal NMPC, the simulation studies demonstrate up to 37.60% and 40.00% error reductions in plunge and pitch motions. Subsequently, we adapt this controller for quadrotor trajectory tracking tasks and compare the results with a benchmark control strategy that loosely coupled the NMPC with INDI. Extensive simulation validations have been performed to track agile trajectories, showing up to a 79.58% reduction in position error and up to a
44.08% reduction in heading error.

The drone market has been steadily growing over the last decades, resulting in drones becoming more and more common 1. For the majority of the time, the drones provide valuable services, from inspection work and deliveries to applications in emergencies, such as search and rescue, or rapid deliveries of medical supplies. Unfortunately, because drones are available to everyone, misuse cannot be fully prevented. They can cause significant disruptions to aviation, violate privacy, or transport illegal substances, leading to financial losses, or more serious consequences. A notable incident, that caused the delay of close to 1,000 flights affecting approximately 140,000 passengers, took place in 2018 at London Gatwick Airport, where 2 drones caused the airport to shut down for over 24 hours [1]. Additionally, it has been reported by Dubai International Airport, that the estimated costs of halting airport operations due to drones resulted in losses of close to $100,000 per minute of downtime2. To counteract these hazardous situations, current anti-drone methods on the market include drone guns, quadcopters using catching systems, and radio frequency jamming systems. However, all of these methods come with their disadvantages. They either involve human interaction, have high operational costs, or cannot intercept quickly over a large area, leaving a large gap in the market for an efficient anti-drone system. The Air-Guard drone introduced in this report aims to fill this market gap, by providing a quick response time to a threat, while not compromising on neutralizing capabilities. It is a fixed-wing drone capable of autonomous visual tracking and catching unlicensed drones via a shooting net mechanism integrated with a parachute. Compared to multi-rotor drones, the Air-Guard drone concept has a longer range and higher efficiency, allowing it to be readily in the air until an unlicensed drone is detected. This is achieved thanks to its unique design inspired by bird morphology. Birds can actively morph their wings and tail surfaces to actively alter their aspect ratio, wing loading, and stability to achieve the most efficient flight configuration over a wide variety of flight profiles. Similarly, the Air-Guard drone morphs its wing and tail such that the drone can be stable while loitering, to then transition into high-g maneuvers in under a second. This allows the drone to more closely follow unlicensed drones in restricted areas and immobilize these drones in a matter of minutes autonomously. The morphing concept of the drone uses multiple actuators and elastic tendons in combination with specially designed artificial feathers to emulate bird morphology. The design also considers previously neglected areas of bird wing anatomy and incorporates bioinspired aerodynamic surfaces to delay stall and increase the maneuverability of the drone. This report is a follow-up of the midterm report, where the general configuration of the drone was set up. It aims to describe the progress on the Air-Guard project, mostly from a technical point of view, but also economic and operational aspects are covered. The UAV design is multidisciplinary and is influenced by many disciplines related to aerospace engineering. Aerodynamics, performance, structures, stability, control, and electronics considerations were combined to make the design possible. It was an iterative process, requiring careful coordination and communication between all the different departments. The result of this work translates into the design of a dual engine, 3.5 kg drone that is capable of loitering for an hour, with a maximum speed of 48 m/s. The wing span in extended configuration is 1.34 m, with a total length of 1.05 m. To neutralize the threat, the Air-Guard chases the unlicensed drone with the help of its high maneuver ability, then fires a net equipped with a parachute from the nose to capture it. The materials used are balsa wood for the fuselage and fixed-wing, while the morphing surfaces are made of aluminum and 3D-printable Celanese VECTRA A950LCP. A great emphasis was put on the sustainability aspect of the drone, which lead to an electrically powered UAV, having a structure that is 99% recyclable.

With the increased focus on sustainability, aircraft are designed to reduce their emissions. One way to accomplish this is by increasing the wing aspect ratio, thereby increasing the aerodynamic efficiency, however this is not without consequences. Increased aspect ratio wings have a higher structural mass, are more susceptible to gust and maneuver loads and generally flutter at lower velocities. Due to the advent of both passive and active control techniques these issues can fortunately be solved by using gust load allevation (GLA), maneuver load alleviation (MLA) and flutter suppression.
Current aeroelastic testing facilities at Delft University of Technology include a gust generator and aeroelastic apparatus, used to suspend a passive wing section in the wind tunnel. The need for ad¬ ditional research on aeroelastic control in order to improve the sustainability and safety of aviation necessitates the development of a new wing section with aileron and spoiler control surfaces that is compatible with current facilities. The development, manufacturing and initial characterization and test¬ ing of this wing section is the subject of the present work.
As the new wing section includes a spoiler, a literature review is performed on this subject. Spoilers function by deflecting into the flow, causing separation aft of the spoiler and creating a large turbulent wake, resulting in a drastic decrease of lift. A linear potential flow model for spoiler aerodynamics developed by Brown and Parkinson was implemented in MATLAB with the intent of implementing this in future aeroelastic models. Verification of this model showed good agreements with original data presented in the paper describing the model.
The passive wing section was chosen as a basis for the new design. The position and size of the control surfaces are determined based on a review of experimental and operational applications. The new wing section was designed, resulting in a self¬contained model, including a single¬board computer, sensors and power supply. Actuation mechanisms were developed for the control surfaces, with a parametric device for control surface free play included in the aileron actuation mechanism. The new wing section was manufactured successfully and control software was implemented using Simulink.
A series of tests were performed to characterize the dynamic behavior of the wing section. Due to a combination of higher inertia and kinematics of the actuation mechanism, the usable bandwidth of the aileron is shown to be lower than that of the spoiler. Aerodynamic results show that the combined use of aileron and spoiler result in a reduction or reversal of the aerodynamic response of the wing. Gust load alleviation results with proportional control show an increase in damping by 1300% and a reduction in peak amplitude of 50% when using the spoiler. Results for the aileron are notice¬ ably less, with a decrease in amplitude of 15% and an increase of damping of 145%. The differences are attributed to both the differences in kinematics of the mechanisms as well as the greater absolute change in lift coefficient obtainable by the spoiler.

Recent trends in aviation highlight the ever-increasing need for fuel economy and sustainability. Active morphing technology can offer significant benefits over conventional wing designs. Inspired by nature, smart morphing technologies enable the aircraft of tomorrow to sense their environment and adapt the shape of their wings in-flight to minimize fuel consumption and emissions. A primary challenge on the road to this future is the question of how to use the knowledge gathered from sensory data to establish an optimal shape
adaptively and continuously in-flight.

To address this challenge, this thesis proposes a novel architecture for online black-box aerodynamic performance optimization for active morphing wings. The proposed method seeks to extend the scope of state-of-the-art online performance optimization methods by integrating a global online-learned radial basis function neural network model with a derivative-free evolutionary optimization strategy. The effectiveness of the optimization strategy was tested on a Vortex Lattice Method aerodynamic model of an over-actuated morphing wing that was corrected using previously collected wind tunnel data. Simulations show that the proposed method is able to control the morphing shape and angle of attack to achieve various target lift coefficients with better aerodynamic efficiency than the unmorphed wing shape. Furthermore, the effectiveness of the optimization architecture was experimentally evaluated on an active trailing-edge camber morphing wing demonstrator with distributed sensing and control, the SmartX-Alpha, in the open jet facility of Delft University of Technology. Compared to the unmorphed shape, a 7.8 % drag reduction was realized, while achieving the required amount of lift. Further data-driven predictions have indicated that even higher reductions in drag are achievable and have provided insight into the trends in optimal wing shapes for a wide range of lift targets.