空中客车

This paper introduces a novel surrogate modeling framework for aerodynamic applications based on Neural Fields. The proposed approach, MARIO (Modulated Aerodynamic Resolution Invariant Operator), addresses non parametric geometric variability through an efficient shape encoding mechanism and exploits the discretizationinvariant nature of Neural Fields. It enables training on significantly downsampled meshes, while maintaining consistent accuracy during full-resolution inference. These properties allow for efficient modeling of diverse flow conditions, while reducing computational cost and memory requirements compared to traditional CFD solvers and existing surrogate methods. The framework is validated on two complementary datasets that reflect industrial constraints. First, the AirfRANS dataset consists of a two-dimensional airfoil benchmark with non-parametric shape variations. Performance evaluation of MARIO on this case demonstrates an order of magnitude improvement in prediction accuracy over existing methods across velocity, pressure, and turbulent viscosity fields, while accurately capturing boundary layer phenomena and aerodynamic coefficients. Second, the NASA Common Research Model features three-dimensional pressure distributions on a full aircraft surface mesh, with parametric control surface deflections. This configuration confirms MARIO's accuracy and scalability. Benchmarking against state-of-the-art methods demonstrates that Neural Field surrogates can provide rapid and accurate aerodynamic predictions under the computational and data limitations characteristic of industrial applications.

We propose $\textbf{FAME}$ (Formal Abstract Minimal Explanations), a new class of abductive explanations grounded in abstract interpretation. FAME is the first method to scale to large neural networks while reducing explanation size. Our main contribution is the design of dedicated perturbation domains that eliminate the need for traversal order. FAME progressively shrinks these domains and leverages LiRPA-based bounds to discard irrelevant features, ultimately converging to a $\textbf{formal abstract minimal explanation}$. To assess explanation quality, we introduce a procedure that measures the worst-case distance between an abstract minimal explanation and a true minimal explanation. This procedure combines adversarial attacks with an optional $VERI{\large X}+$ refinement step. We benchmark FAME against $VERI{\large X}+$ and demonstrate consistent gains in both explanation size and runtime on medium- to large-scale neural networks.

Technological advancements in human-machine interfaces increasingly confine human operators to the role of passive supervisors. Under such conditions, the out-of-the-loop phenomenon, driven by vigilance decline and task disengagement, can result in degraded performance, when automated systems fail unexpectedly. Despite decades of research on the topic, characterizing, quantifying and predicting these performance problems remain difficult. This study aimed at identifying distinct oculomotor signatures associated with task disengagement and vigilance reduction during a specifically designed supervisory task. Participants viewed a simplified aircraft interface and were asked to monitor an autopilot system and validate or reject its decisions. Task disengagement was manipulated by varying the time delay before participants were required to intervene: in some cases, responses were frequent, while in others participants had to wait longer before having to intervene. Vigilance was manipulated through time-on- task. Results showed that participants' performance was negatively impacted by task disengagement (slower reaction times when less solicited). Although visual exploration decreased both under low vigilance and low engagement, other oculomotor markers (i.e., saccades, fixations, eyelid opening and blinks) could dissociate the two states. Finally, logistic classifiers revealed that specific oculomotor features predicted participants' reaction times above chance. In summary, this study identified distinct oculomotor patterns linked to vigilance decline and task disengagement, two factors contributing to the out-of-the-loop phenomenon, which were also predictive of individuals' performance. These results are not only relevant for the monitoring of piloting activities, but contribute to the general knowledge of the impact of task disengagement on human performance.