The actual performance of model-based path-following methods for Unmanned Aerial Vehicles (UAVs) show considerable dependence on the wind knowledge and on the fidelity of the dynamic model used for design. This work analyzes and demonstrates the performance of an adaptive Vector Field (VF) control law which can compensate for the lack of knowledge of the wind vector and for the presence of unmodelled course angle dynamics. Extensive simulation experiments, calibrated on a commercial fixed-wing UAV and proven to be realistic, show that the new VF method can better cope with uncertainties than its standard version. In fact, while the standard VF approach works perfectly for ideal first-order course angle dynamics (and perfect knowledge of the wind vector), its performance degrades in the presence of unknown wind or unmodelled course angle dynamics. On the other hand, the estimation mechanism of the proposed adaptive VF effectively compensates for wind uncertainty and unmodelled dynamics, sensibly reducing the path-following error as compared to the standard VF.

The high maneuverability of fixed-wing unmanned aerial vehicles (UAVs) exposes these systems to several dynamical and parametric uncertainties, severely affecting the fidelity of modeling and causing limited guidance autonomy. This article shows enhanced autonomy via adaptation mechanisms embedded in the guidance law: a vector-field method is proposed that does not require a priori knowledge of the UAV course time constant, coupling effects, and wind amplitude/direction. Stability and performance are assessed using the Lyapunov theory. The method is tested on software-in-the loop and hardware-in-the-loop UAV platforms, showing that the proposed guidance law outperforms state-of-the-art guidance controllers and standard vector-field approaches in the presence of significant uncertainty.

This paper investigates the use of structured robust techniques for the control design of the aerodynamic descent phase of a reusable rocket. A unified control strategy based on the use of linear-time-invariant structured [ ] controllers for both attitude and position is proposed and verified in both frequency and time domains. The architecture of the controller and the observed performance are discussed. Moreover, the nonlinear behavior of the control strategy is analyzed by looking at both the gain-scheduling law in conditions of plant-controller misalignment through the use of misalignment matrices, and simulation results obtained with a high-fidelity, nonlinear six-degree-of-freedom closed-loop model in perturbed conditions. All the results are based on Cooperative Action Leading to Launcher Innovation in Stage Tossback Operations (CALLISTO), a reusable rocket demonstrator jointly developed by German Aerospace Center (DLR), Japan Aerospace Exploration Agency (JAXA), and National Centre for Space Studies (CNES).

Reusable rockets must rely on well-designed Guidance, Navigation and Control (GNC) algorithms. Because they are tested and verified in closed-loop, high-fidelity simulators, emphasizing the strategy to achieve such advanced models is of paramount importance. A wide spectrum of complex dynamic behaviors and their cross-couplings must be captured to achieve sufficiently representative simulations, hence a better assessment of the GNC performance and robustness. This paper focuses on of the main aspects related to the physical (acausal) modeling of reusable rockets, and the integration of these models into a suitable simulation framework oriented towards GNC Validation and Verification (V&V). Firstly, the modeling challenges and the need for physical multibody models are explained. Then, the Vertical Landing Vehicles Library (VLVLib), a Modelica-based library for the physical modeling and simulation of reusable rocket dynamics, is introduced. The VLVLib is built on specific principles that enable quick adaptations to vehicle changes and the introduction of new features during the design process, thereby enhancing project efficiency and reducing costs. Throughout the paper, we explain how these features allow for the rapid development of complex vehicle simulation models by adjusting the selected dynamic effects or changing their fidelity levels. Since the GNC algorithms are normally tested in Simulink^®, we show how simulation models with a desired fidelity level can be developed, embedded and simulated within the Simulink^® environment. Secondly, this work details the modeling aspects of four relevant vehicle dynamics: propellant sloshing, Thrust Vector Control (TVC), landing legs deployment and touchdown. The CALLISTO reusable rocket is taken as study case: representative simulation results are shown and analyzed to highlight the impact of the higher-fidelity models in comparison with a rigid-body model assumption.