Fault-tolerant flight control remains a major challenge as aircraft systems become increasingly autonomous and must operate under uncertain conditions and potential actuator failures. Reinforcement learning has shown strong potential for learning control policies directly from interaction with the environment, but purely offline-trained agents often lack the ability to adapt once deployed.

This thesis proposes a hybrid reinforcement learning framework that combines offline deep reinforcement learning with online adaptive control. A novel RUN-DSAC-IDHP controller is developed, integrating an uncertainty-aware offline policy learned with RUN-DSAC with an online Incremental Dual Heuristic Programming (IDHP) adaptation layer. The offline component provides a high-performance baseline policy, while the IDHP actor continuously adapts the control policy online when the aircraft dynamics change.

The proposed approach demonstrates how combining offline deep reinforcement learning with online adaptive control enables controllers to maintain strong baseline performance while adapting in real time to faults and changing flight conditions. ... Read more

Smart Vortex Generators (SVGs) are used in a CFD study for control of loads on aeroelastic wings governed by mass-damper-spring structural model in the presence of unsteady wind gust disturbances, leveraging interpretable reduced-order system identification. The primary objective of the study is thus gust load alleviation (GLA). Unlike prior formulations, the dominant disturbance, such as a wind gust, is included explicitly as a control input within the model, allowing the MPC controller to respond adaptively and in anticipation to external system forcing. System dynamics are identified partly by Sparse Identification of Nonlinear Dynamics with Control (SINDYc) and Linear Parameter-Varying (LPV) system. SINDYc captures the wing-gust subsystem, while LPV offers physics-interpretable and sparse representation of the nonlinear aerodynamic effects associated with the different operating regimes of the SVGs—vortex generator and massive flow separation. Two types of nonlinearity are observed. Firstly, nonlinearities affecting amplitude and non-minimum-phase-like behaviour of the induced loads can be captured through one scheduling parameter—the SVG deflection angle. The second type of identified nonlinearity concerns the different GLA capability of SVGs in the presence of gusts with varying magnitude. This can be captured by second scheduling parameter—the SVG deflection angle multiplied by the effective angle of attack, identified by SINDYc. Once fully coupled reduced-order fluid-structure interaction dynamical model is constructed, model-predictive control (MPC) is deployed in order to study the maximum potential GLA capability of SVGs. It has been shown in previous studies that simple on/off control strategy induces secondary structural oscillations, resulting in more structural fatigue. It is shown here that optimal control strategy can mitigate these to an extent. MPC-controlled SVGs can best alleviate loads of low frequency high amplitude gusts, by up to ΔCL ≈ 0.23, while high frequency gusts require head start for SVG actuation for optimal GLA performance. In summary, the findings of this thesis demonstrate that SVGs constitute a viable candidate technology for active gust load alleviation when deployed together with optimal control framework.
... Read more

Expanding the operational capabilities of Micro Air Vehicles (MAVs) hinges on control systems that manage highly nonlinear dynamics across broad flight envelopes. Incremental Nonlinear Dynamic Inversion (INDI) is popular for its simplicity and modest modeling needs, but its assumption of infinitely fast actuators and neglect of state-dependent effects limit performance when actuators have slow or heterogeneous dynamics or when aerodynamic effects are significant. Actuator Nonlinear Dynamic Inversion (ANDI) overcomes these limitations by explicitly incorporating state-dependent dynamics and finite actuator bandwidth into the control law, enabling improved tracking performance across diverse actuator configurations. This work implements the full ANDI stabilization controller on the Cyclone, a hybrid MAV tail-sitter, using cascaded complementary filtering for state estimation. Simulation and flight experiments validate the approach and assess whether this compensation yields practical performance gains, establishing ANDI as a viable, generic control solution for MAVs
Code is available at:
https://github.com/tudelft/paparazzi/tree/feat_stabilization_andi_controller. ... Read more

Delay propagation is a significant driver of flight delay in aviation networks, yet modelling it at a network-wide scale remains challenging. This study investigates to what extent scheduled max-plus linear systems, as used in railway delay modelling, can be applied to aviation networks. Using the Hawaiian Airlines network as a case study, a methodology is developed to model aircraft rotation and passenger transfer precedence relations within a max-plus linear system. The approach enables the calculation of stability indicators such as maximum cycle mean, recovery times, and network slack, as well as the simulation of delay propagation under various initial delay scenarios.

Results show that the recovery matrix is a valuable tool for identifying structurally vulnerable parts of the network and for assessing the impact of holding aircraft for transferring passengers. However, predictive accuracy of delay propagation for individual flights is limited, primarily due to uncertainties in process time estimation and incomplete knowledge of precedence relations. The 24-hour periodicity of aviation timetables, combined with large overnight buffers, further limits multi-day delay propagation modelling. These limitations are partly specific to the case under study and partly inherent to the deterministic, periodic structure of scheduled max-plus systems.

The study concludes that max-plus linear systems can provide meaningful insights into structural robustness and the systemic impact of schedule design choices, but their use for precise short-term delay prediction in aviation is constrained without high-quality operational data. Future work should explore integration of stochastic max-plus models, application to networks with shorter periodicity, and validation using airline-provided operational datasets. ... Read more

The Flying-V is a tailless, V-shaped flying-wing type aircraft that promises to offer significant increases in aerodynamic efficiency. Due to its configuration, the Flying-V faces some control and stability related issues. These include limited control authority, pitch break tendencies and non-ideal handling qualities. To enhance the handling qualities of the Flying-V, Incremental Nonlinear Dynamic Inversion (INDI)-based flight control systems have been proposed. INDI, a sensor-based alternative to conventional Nonlinear Dynamic Inversion (NDI), is rooted in the principle of feedback linearization. Unlike NDI, INDI does not depend heavily on accurate on-board models (OBM), thereby offering increased robustness to aerodynamic uncertainties. However, singular perturbations—such as time delays, aeroelastic effects, and additional unmodeled or unknown dynamics—have been identified as challenges for INDI-based control laws. Various strategies have been explored to improve the overall robustness of INDI-based flight control systems, including outer-loop tuning and inversion loop augmentation strategies. In this research a multi-loop µ-optimal approach for designing robust inversion-based flight control laws is explored for the design of an explicit model-following pitch-rate control system for a short-period approximation of the Flying-V’s longitudinal dynamics. The design problem takes into account both regular and singular perturbations. To assess the robust stability and performance of the proposed control systems, a structured singular value analysis was performed. It was concluded that a multi-loop synthesis approach is capable of achieving better robust stability and performance levels when compared to either strictly inner-loop or outer-loop synthesis. As such, it can be concluded that multi-loop synthesis approaches are best capable of leveraging the robustness functionalities of multi-loop inversion-based control systems. ... Read more