Engineering

Abstract: This paper presents the development and implementation of a psychological model aimed at predicting the mental states of air traffic controllers (ATCOs). The proposed model aims to advance human–machine collaboration in air traffic management by enabling the precise prediction of critical operator cognitive and affective states, including mental workload, fatigue, stress, and attentional engagement. By formally integrating core cognitive processes—namely perception, comprehension, and decision-making—within its architecture, the model provides a principled framework for the continuous monitoring and real-time adaptation of support systems. Such adaptive capabilities are intended to optimize the allocation of assistance provided by artificial agents, thereby strengthening human–system coordination and contributing to enhanced operational safety and efficiency within the complex and highly dynamic environment of air traffic control. To estimate the parameters of the model, several air traffic simulations were conducted with expert controllers. In these simulations changes to traffic situations were introduced. Those changes could affect the controllers' mental states. The results of these changes were observed in the measured verbal and psychophysiological dependent variables. This paper presents results that partially validate the initial parameters of the models. These results will contribute to a future improvement of the model by refining the parameters of the proposed formulas for calculating mental workload, fatigue, stress, and vigilance in the air traffic control task.

Abstract: When general-aviation aircraft leave a training airspace and several return to the circuit together, the tower controller must assign one of six allocation schemes to each aircraft within seconds. Historical records, however, resolve identical situations in different ways, so a purely supervised classifier trained on them saturates: on the SR-20 sample used here (= 20,000, six schemes) it reaches only 87.69 % accuracy and 86.85 % macro-F1. This paper moves the supervision target from the records to an explicit, auditable matching-rule model that encodes the operating regulations as a ternary structure with three-source evidence and a trend-strength gate; only its continuous coefficients are estimated by a conditional-logit fit. The rule posterior is distilled into a LightGBM student as a soft label. Against the deterministic rule standard of the held-out test split the student reaches 95.48 % accuracy, 94.99 % macro-precision and 95.04 % macro-F1, i.e., meets deployment requirements. The single-modality controls fail - a BiLSTM’s macro-F1 on the flight frames reaches 81.94 % and a Chinese-RoBERTa text classifier’s macro-F1 reaches 50.48 % - so the two modalities carry complementary information; a random forest’s macro-F1 reaches 90.89 % but is less auditable. Because every training label is generated by a named rule, the deployed matcher stays traceable, and retraining costs under one minute on a desktop CPU, which is the property a safety-critical tower function requires.

Abstract: Reinforcement learning (RL) is increasingly being investigated for spacecraft guidance, navigation, control, and mission autonomy, particularly where uncertainty, nonlinear dynamics, limited onboard computation, or changing operating conditions challenge conventional approaches. This review examines RL and closely related learning-based methods across major space-control applications, including attitude and orbit control, low-thrust and cislunar guidance, planetary landing, small-body operations, rendezvous and docking, space robotics, surface mobility, and mission-level autonomy. Reported results are assessed with respect to classical baselines, quantitative performance, robustness, computational cost, and validation fidelity from simulation to hardware and flight demonstrations. Across the literature, learned controllers often show advantages in adaptation, robustness to modeled uncertainty, and rapid onboard evaluation, while numerical optimization and classical control generally retain advantages in nominal optimality, interpretability, and established guarantees. Hybrid architectures, in which learning modifies or augments a conventional guidance or control law, provide some of the most consistent results. Important limitations remain in out-of-distribution generalization, benchmarking, statistical reporting, navigation realism, verification, implementation assurance, and independent high-fidelity validation. The review concludes with proposed benchmark problems, reporting practices, and a staged validation pathway for assessing learned controllers in increasingly realistic space-system environments.

Abstract: A universal spatiotemporal coupling optimal low-dimensional dynamical system algorithm for compressible turbulence which unifies RANS, LES and DNS in the sense of Box filtering/spatial grid averaging and large-scale grids: the LMS method is outlined first. Then the equation error coursed by finite truncation in computational fluid dynamics(CFD) is reported first time. The use of the LMS method can reduce this error, leading to a unified new paradigm in CFD that combines numerical simulation of turbulence with computational fluid dynamics. Finally, a major surprising discovery: The essences of turbulence modelling theory is nothing, but just a part of the equation error coursed by finite truncation in CFD.

Abstract: The increasing use of lightweight, large compliant thin-walled structures (LCTS) made of carbon-fiber-reinforced polymer (CFRP) in aircraft fuselage assemblies contributes significantly to fuel efficiency and emission reduction. However, the low bending stiffness and large dimensions of these components make them highly susceptible to elastic deformation under gravity, fixture contact forces, and handling loads, so the geometry observed during manufacturing, inspection, and assembly does not represent the intrinsic, stress-free shape of the component. Because this gravity-free shape cannot be measured directly, it must instead be approximated from measurements taken under known loading conditions, combined with a numerical model of the component's structural response -- a separation that is required for reliable geometric inspection, digital assembly simulation, variation analysis, and fixture design. Minimal-contact supports commonly used for this purpose cause excessive deflection in structures of this size, while conventional multi-point supports leave the force distribution at each support point unknown, introducing ambiguity into the gravity compensation. This work addresses both limitations with a smart fixture system whose instrumented actuators render the imposed displacements and support states fully known, enabling a controlled, over-constrained support condition without loss of traceability of the applied loads. Surface measurements acquired on this fixture are combined with finite element simulations of the same loading configurations to approximate the gravity-free shape of large compliant CFRP structures, isolating manufacturing deviations from gravity-induced elastic deformation and yielding a gravity-free description of the component suitable for downstream geometric evaluation.

Abstract: The benefit of exhaust heat recuperation in a micro turbojet depends on the fuel saved at the required thrust and on recuperator mass and pressure losses. A 1000 N engine with a semicircular-channel matrix is analysed in ASTRA at constant static thrust for temperature recovery ratios θ = 0–0.7. Compressor and turbine characteristics are interpolated, scaled and extrapolated during component matching. Original, half and zero pressure-loss cases retain the same matrix mass. With the original losses, the lowest calculated engine-plus-fuel mass for one hour occurs at θ = 0.3, but the compressor stability margin of 5.856% is below the adopted 10% threshold. Among points meeting this thresh-old, θ = 0.2 leaves 1.565 kg for additional installation mass and requires a turbine inlet temperature (TIT) of 1234.38 K, 65.71 K above the reference. For one hour with 90% nominal operation and no fuel-flow difference during the remaining time, the allowance falls to 0.962 kg. Half-loss and ze-ro-loss cases give one-hour allowances of 4.198 and 7.716 kg. These sensitivities define targets for recuperator design; the mass, thermal strength and life of the complete installed assembly remain to be established.

Abstract: Systematic comparisons of the classic airfoil literature with modern codes have been limited less by physics than by the labour of extracting data from scanned charts. This study shows that optical character recognition and an AI agent, working under the author’s direction, can automate this chain: source retrieval, chart calibration, visual data reading, simulation campaigns, quality control and documentation. With this workflow, all 112 section charts of NACA Report 824 and three planform figures of Hoerner’s Fluid-Dynamic Lift were recomputed with XFOIL, MSES and AVL in about 7.5 hours of elapsed time. Lift-curve slopes agree to a median +2.2 % (XFOIL), and the overprediction grows with thickness. Zero-lift angles are too negative in proportion to camber in both codes; transition modelling and tunnel-wall corrections were ruled out as causes. Drag is underpredicted by about 15 % at all Reynolds numbers. The offset vanishes for N_crit ≈ 5, an effective parameter for surface imperfections or the shared boundary-layer model. Maximum lift is overpredicted, most for thin sections. AVL reproduces Hoerner’s planform data within a few per cent where the flow is attached and the edges are sharp. Automated checks found errors in the source, the codes and the agent’s own work; human verification remains essential.

Abstract: This paper presents a thermodynamic and fluid-dynamic evaluation of superfluid helium (He-II) at 2.17 K within a non-chemical electromagnetic propulsion framework (AETHER). By coupling vortex-stabilized cryogenic expulsion with pulse-based thrust control, the architecture bypasses chemical combustion, utilizing He-II for high-flux thermal management while generating directional thrust via high-frequency electromagnetic acceleration of ionized helium plasma.

Abstract: Background-oriented schlieren (BOS) is an optical method for measuring refractive dis- 13 placement without inserting a probe into the flow. Fourier-based methods such as fast 14 checker demodulation (FCD) can be affected by background occlusions and may produce 15 large displacement artifacts around obstructed regions or sharp displacement boundaries. 16 Correlation-based methods are generally more robust to local image disturbances, but 17 their spatial resolution is limited by the interrogation-window size. Here, we propose 18 stripe-decomposition BOS (SD-BOS), which extends GPU-accelerated Simplified BOS (S- 19 BOS) to two-component displacement measurement using a monochrome checkerboard. 20 SD-BOS separates the image into two orthogonal stripe carriers and calculates their phase 21 differences independently. Edge-adaptive blending and confidence-based occlusion han- 22 dling are used to reduce artifacts near sharp gradients and background obstructions. In 23 synthetic tests, SD-BOS reduced vector RMSE by 29.8% for a smooth Gaussian field and 24 21.9% for a shock-like field relative to FCD. For an opaque circular obstruction, the 95th- 25 percentile displacement error outside the detected mask was 86.4% lower than for FCD; 26 vectors within the mask were excluded. In wind-tunnel images, SD-BOS showed lower 27 image-space high-pass fluctuations than FCD, but no independent reference displacement 28 field was available to assess absolute accuracy. The complete SD-BOS pipeline processed 29 512 × 512-pixel images at 115.3 frames per second (fps) on an NVIDIA GeForce RTX 4070 30 SUPER GPU, approximately 23 times the CPU rate measured on an AMD Ryzen 7 5700X. 31 These results indicate improved robustness in the evaluated synthetic occlusion case and 32 highlight the need for independent validation of wind-tunnel measurements.

Abstract: Strong tip vortices generated by hovering rotors contribute to induced aerodynamic losses. Blade-tip jet injection offers a means of modifying the tip-vortex system and the associated aerodynamic loads. In this study, a computational model of a hovering rotor was developed using the multiple reference frame (MRF) approach and validated against the experimental measurements of the Caradonna–Tung rotor. At a representative high-tip-speed condition, the effects of jet velocities equal to one, two, and three times the blade-tip speed, as well as the effect of jet-orifice diameter, were investigated in terms of rotor thrust, torque, and tip-vortex structure. The results indicate that a jet velocity of 200 m/s (twice the blade-tip speed) yields the greatest thrust increase among the investigated jet-velocity cases: thrust increases by 5.43%, accompanied by a 4.17% increase in torque. Increasing the orifice diameter within the investigated range further increases both thrust and torque. Flow-field visualizations suggest that jet-induced momentum injection modifies tip-vortex coherence, wake organization, and local flow structures. These findings demonstrate the potential of distributed blade-tip injection for controlling hovering-rotor aerodynamics while highlighting the accompanying torque penalty.

Abstract: The paper deals with the development of a model-based optimization framework for the design of power management strategies of a lightweight compound helicopter UAV equipped with hybrid propulsion, with focus on the transition manoeuvres from cruise to hover. A high-fidelity multi-physics model of the hybrid propulsion system, composed of a combustion-engine connected to a nose propeller for forward flight phases, and two permanent magnets synchronous machines driving variable-pitch main and tail rotors to provide control during vertical flight phases, is firstly presented. In the model, the aerodynamic loads are obtained by applying an advanced BEMT-based code, including post-stall, tip-loss and low-Reynolds corrections, as well as QPROP-derived induced velocities. The dynamics of the electric machines is simulated via equivalent bi-phase model taking account of the Field-Oriented Control technique used for the three-phase electronic commutation. As a first step of the dynamic optimization process, a so-called fixed-point optimization is obtained, by finding the speeds and blade pitch angles of the rotors that minimize the steady-state electrical power absorption in vertical flight phases. To compare the optimization results with a baseline, a reference manoeuvre is then defined, in which the rotors speeds are ramped from standstill to the optimal fixed-point values, while the blade pitch angles are dynamically adjusted to ensure the UAV vertical equilibrium. Subsequently, a constrained optimization problem is applied by using the high-fidelity model, determining the main rotor commands that minimize the power consumption during transition. Finally, the robustness of the system performances against parameter uncertainties is statistically characterized via Monte Carlo simulations. The results demonstrate the effectiveness of the proposed strategies in reducing the energy consumption and in improving the overall performances with respect to a baseline manoeuvre, highlighting the effectiveness of model-based optimization for the power management design of novel UAV platforms.

Abstract: Onboard processing for remote sensing satellites has long been of interest in ocean monitoring because it can provide timely information on rapidly changing ocean environments. Recently, the integration of artificial intelligence (AI) into onboard processing has emerged as an active research topic. However, application scenarios and on-orbit demonstrations of AI-based onboard processing remain limited. In this study, we achieved the first on-orbit detection of ocean eddies using AI on the Haishao-1 (HS-1) spaceborne synthetic aperture radar (SAR) satellite, marking the first attempt to monitor an ocean dynamic process directly onboard. Specifically, an onboard application module was developed for HS-1, integrating image preprocessing and deep learning-based detection based on our previously developed Eddy Detection based on YOLO (EOLO) model. On-orbit experiments verified stable execution of the application module. Ground reproduction of the tasks using hardware identical to the onboard computing units yielded an average processing time of 3.4-6.1 s per image. Evaluation against expert annotations of 288 HS-1 scenes showed that EOLO, originally developed using Sentinel-1 imagery, achieved an F1 score of 78% after threshold adjustment. The application module has also been deployed on the Haishao-2 platform for Arctic sea ice extraction, further demonstrating its extensibility. This study presents the implementation and on-orbit validation of the onboard application module, providing a practical reference for future applications.

Abstract: Embedding the engines in the rear fuselage is a promising concept to reduce aircraft fuel consumption, but it leads to complex load paths and can introduce engine vibrations into the fuselage skin. Sandwich panels with an auxetic metamaterial core are a potential skin concept for this region, as they can be tailored locally and enhance energy dissipation and passive vibration damping. However, modeling the individual cells for a fuselage structure leads to infeasibly large system sizes. This work therefore presents a scale-bridging analysis framework that homogenizes sandwich materials with auxetic cores into shell stiffnesses for macroscale structural design. For this, a first-order computational homoge-nization of the core combined with classical laminate theory and an asymptotic homogenization of the complete sandwich section are considered. Both are compared against a reference model with a resolved core under tension, shear and bending as well as in a modal analysis, with deviations below 6% for the asymptotic homogenization. The homogenized stiffnesses are then used in an isogeometric analysis of a rear fuselage with embedded engines. Under a 2.5 g pull-up maneuver, the core topology and its orientation directly influence the global deformation, making the core topology a design variable at the aircraft scale.

Abstract: Predicting the aerodynamic performance of multirotor unmanned aerial vehicles (UAVs) remains challenging because of complex rotor–rotor and rotor–body interactions. Experimental characterisation of these interactions is often costly and time-consuming, making computational methods valuable for aerodynamic design and optimisation. This study presents a coupled computational fluid dynamics and blade element momentum theory (CFD-BEMT) framework for evaluating a multirotor UAV developed for power-line inspection. The airflow was simulated using OpenFOAM, while the propellers were represented as virtual rotors through distributed momentum-source terms calculated using BEMT. The CFD methodology was assessed by comparing predicted Ahmed-body drag coefficients at different yaw angles with experimental measurements. Steady-state Reynolds-averaged Navier–Stokes simulations provided satisfactory predictions at low yaw angles, while unsteady simulations generally improved the predictions at higher yaw angles, where large-scale flow separation and wake unsteadiness became significant. Standalone static-thrust measurements were also used as reference data to examine changes in rotor loading after integration with the UAV. The computational framework was subsequently applied to a baseline bluff-body configuration, V01, and a redesigned configuration, V02, incorporating smoother fuselage contours, improved payload integration, and modified rotor placement. A comprehensive simulation matrix was used to investigate the effects of forward velocity, propeller rotational speed, and nose-down angle of attack. The results show that propeller operation substantially modifies the aerodynamic response through nonuniform rotor-induced flow and rotor–body interactions. Compared with V01, the redesigned V02 configuration generally exhibits smaller variations in lift, drag, and lift-to-drag ratio across the investigated angles of attack. However, V02 does not provide uniformly lower drag or a consistently higher lift-to-drag ratio under all operating conditions. Its principal advantage is a more consistent aerodynamic response and reduced sensitivity to changes in nose-down angle of attack. The results demonstrate the capability of the coupled CFD-BEMT framework to assess installation effects, quantify rotor–body interactions, and support the aerodynamic development of multirotor inspection UAVs.

Abstract: This study empirically evaluates deep learning–based image analysis systems using real-world UAV imagery and investigates their applicability to public safety surveillance and enforcement. Field experiments were conducted with a police-operated mission drone and a high-resolution commercial UAV platform, covering daytime electro-optical (EO) and nighttime infrared (IR) imaging across varying altitudes and observation distances. Detection performance was assessed via valid detection ratio, false positive rate, and effective detection range. Results show that detection reliability is strongly constrained by input image resolution, target spatial scale, and environmental conditions, even when identical deep learning algorithms are applied. The mission drone maintained useful performance at short range and low altitude but degraded rapidly at medium and long ranges, whereas the high-resolution platform sustained stable detection over extended distances. Using ground sampling distance (GSD)-based optical equivalence, UAV imaging parameters were mapped to equivalent ground-based configurations, yielding conceptually derived performance thresholds linking image resolution and observation distance to detection reliability. These thresholds are interpreted in relation to ground-based intelligent traffic enforcement requirements, though empirical validation under fixed-camera conditions remains for future research. These findings highlight that minimum system-level imaging requirements, rather than algorithmic accuracy alone, are essential for designing, procuring, and certifying public-sector video analytics systems.

Abstract: Flight-control laws are assessed by their frequency response, stability margins, gust rejection and behaviour under failure; connectome-derived controllers have so far been shown mainly in time-domain demonstrations. We applied these tests to a spiking model of the fruit fly's visual yaw pathway, a 20,556-neuron FlyWire subcircuit from the T4/T5 motion detectors to the descending neuron DNa02, with the aircraft's yaw rate imposed directly as the firing rate of artificial T4/T5 sources, and compared it with rewired copies and a classical yaw damper. Under sinusoidal yaw, the right-minus-left DNa02 firing rate followed the yaw rate with coherence 0.98 at 1 Hz and with the stabilizing sign. It ranked first against 19 degree-preserving rewirings (median 0.31; p = 0.05) and narrowly first against 19 cell-type-preserving rewirings, which leave every input to the HS, VS, H2 and DNa02 cells unchanged. Closed around the rudder of a simulated six-degree-of-freedom Cessna 172 in moderate turbulence and tuned under the same margin rule as the yaw damper, it reduced RMS yaw rate by 19%, against 57% for the damper. Of 11 wirings it had the lowest RMS yaw rate and the lowest RMS sideslip on both aircraft models. The three rewired controllers that came closest on yaw rate in the Cessna model were saturated relays flying skidding turns. No lesion-study flight departed, so the pre-specified departure test could not separate the wirings; under random neuron loss the real wiring's mean RMS yaw rate rose toward the bare airframe's but not above it, partly because the loss often silenced its one active readout neuron, and silencing its six HS cells returned it to the bare airframe's yaw rate. Flight-control tests can therefore measure what a connectome's wiring contributes to a controller, and they expose failure modes, such as relay-like skidding, that a single performance score hides.

Abstract: Space debris fragments too small to track can still reach speeds capable of disabling a spacecraft. The number of such fragments remains uncertain, making it difficult to assess risks to satellites supporting Earth-critical services such as broadband and GPS. We combine radar detection with a Bayesian model to adaptively choose where a radar should point next, aiming to reduce uncertainty. We analyzed about 24 hours of data from each of the two European Incoherent Scatter Scientific Association (EISCAT) radars: the Tromsø ultra high frequency radar and the EISCAT Svalbard radar. We first used a Generalized Matched Filter to search for radar signals consistent with moving objects. Tests with 4,000 simulated object signals in raw voltage data led us to select cell-averaging Constant False Alarm Rate (CA-CFAR) detection. For external validation on another radar, we applied the CA-CFAR detector selected using the Tromsø recordings to the Svalbard recordings without retuning. The method found 155 events at Tromsø and 222 at Svalbard. Of these, 106 and 119, respectively, could not be confidently matched to cataloged objects. We used the Tromsø rate of unresolved events to calibrate a Bayesian population model informed by the Orbital Debris Engineering Model (ORDEM). With this model, we compared radar pointing schedules across 300 simulated worlds, giving each schedule 24 observations. Adaptive pointing with service weights reduced uncertainty at the final observation by 41.2%, 15.4%, and 5.35% relative to fixed, random, and preplanned schedules, respectively. The additional improvement over unweighted adaptive pointing was estimated at 1.55% (95% interval: -0.47% to 3.53%), a small difference that was not statistically significant. The simulations provide strong evidence that adaptive pointing can reduce uncertainty around hidden debris using the same number of observations, potentially improving the information used to assess risks to important satellite services.

Abstract: This paper presents the end-to-end multidisciplinary development and flight-test assessment of a Class I fixed-wing mini unmanned aerial vehicle (UAV) designed for surveillance and reconnaissance missions. The development process progresses from mission and system-level requirements through conceptual, preliminary, and detailed design, followed by prototype fabrication and experimental flight testing. Conceptual design establishes aircraft mass, wing and power loading, aerodynamic configuration, wing geometry, airfoil selection, mass distribution, and initial stability characteristics. The preliminary and detailed stages progressively increase analysis fidelity through assessments of aerodynamics, structures, aeroelasticity, propulsion, stability and control, performance, and thermal management. High-fidelity Computational Fluid Dynamics (CFD) is employed for aerodynamic char-acterization and load prediction, while Finite-Element (FE) and aeroelastic models are used to assess strength, stiffness, static and dynamic aeroelastic behavior. The aircraft configuration incorporates a V-tail, electric pusher propulsion, composite structural components, integrated control surfaces, wingtip devices, and dedicated cooling arrangements. A representative 3D-printed flight-test article retaining the principal aerodynamic architecture was subsequently evaluated in climb, straight-and-level flight, sustained turns, and low-speed operation. The vehicle exhibited stable and predictable handling throughout the tested envelope. Flight-derived aerodynamic characteristics showed close agreement with the detailed-design CFD database, with drag-coefficient deviations remaining below approximately 3% in straight-and-level flight and within approximately 1.5% during climb. Larger discrepancies, reaching approximately 16%, were observed during sustained turns, highlighting the increased uncertainty associated with maneuvering flight and with drag values inferred using the propulsion model. Overall, the study demonstrates a traceable progressive-fidelity methodology for advancing a mini-UAV from requirements and conceptual sizing to multidisciplinary verification and prototype-level flight assessment.

Abstract: Tilt-wing electric vertical take-off and landing (eVTOL) aircraft face distinctive safety challenges during transition due to configuration variations, highly nonlinear aerodynamics and coupled effects of distributed electric propulsion. These characteristics impose substantial design efforts when conventional envelope protection methods are applied. This paper proposes a transition-corridor-based protected control framework which is compatible with both nonlinear model-based control and conventional feedback control. First, a transition corridor prediction method incorporating electric propulsion constraints is presented, yielding pitch angle and pitch angular acceleration limits as limit commands. Second, a corridor-based protected control strategy is established and integrated into incremental nonlinear dynamic inversion (INDI) and proportional-integral-derivative (PID) controllers. Furthermore, a comparative study is conducted across different protection strategies and controllers to assess the proposed protected control method in safety-critical transition flight. Simulation results demonstrate that the proposed method provides effective protection while exploiting available maneuvering margins. Moreover, incorporating electric propulsion constraints improves the physical consistency of corridor prediction and envelope protection. Overall, this study provides a practical corridor-based protection framework that integrates transition corridor constraints into closed-loop flight control, enabling safer transition for tilt-wing eVTOL aircraft.

Abstract: The management of NOTAMs on the flight deck remains a challenge for flight crews, who must process large volumes of dense textual information to identify the notices relevant to their mission. Despite ongoing modernisation efforts, the NOTAM system still relies on a legacy format with limited support for automated filtering and prioritisation, which increases pilot workload and motivates the use of Natural Language Processing techniques for information management. This paper presents an NLP framework based on a shared pre-trained BERT encoder with two task-specific prediction heads, designed to support NOTAM analysis at different stages of the flight. The first task classifies each NOTAM into one of 20 operational categories to support pre-flight briefing. The second task estimates the operational severity of a NOTAM under three emergency diversion scenarios to support in-flight decision-making within cockpit decision-support systems. To address the lack of annotated data and the imbalance across categories, a dataset of 3,000 NOTAMs was built by combining expert annotation with synthetic samples generated under human supervision. The proposed framework, based on a DistilBERT encoder fine-tuned with LoRA, achieves a macro F1-score of 0.97 on classification and R² values at or above 0.80 across the three severity scenarios.