Portail HAL ONERA
Not a member yet
12842 research outputs found
Sort by
De l’imagerie spatiale à aéroportée : synthèse d’images SAR par modèles de fondation pour une adaptation multi-échelle
No full textInternational audienceThe availability of Synthetic Aperture Radar (SAR) satellite imagery has increased considerably in recent years, with datasets commercially available. However, the acquisition of highresolution SAR images in airborne configurations, remains costly and limited. Thus, the lack of open source, well-labeled, or easily exploitable SAR text-image datasets is a barrier to the use of existing foundation models in remote sensing applications. In this context, synthetic image generation is a promising solution to augment this scarce data, enabling a broader range of applications. Leveraging over 15 years of ONERA’s extensive archival airborn data from acquisition campaigns, we created a comprehensive training dataset of 110 thousands SAR images to exploit a 3.5 billion parameters pre-trained latent diffusion model [1]. In this work, we present a novel approach utilizing spatial conditioning techniques within a foundation model to transform satellite SAR imagery into airborne SAR representations. Additionally, we demonstrate that our pipeline is effective for bridging the realism of simulated images generated by ONERA’s physicsbased simulator EMPRISE [2]. Our method explores a key application of AI in advancing SAR imaging technology. To the best of our knowledge, we are the first to introduce this approach in the literature.La disponibilité des images satellites issues du radar à ouverture synthétique (SAR) a considérablement augmenté ces dernières années, avec des jeux de données désormais accessibles commercialement. Cependant, l’acquisition d’images SAR haute résolution en configuration aéroportée reste coûteuse et limitée. Ainsi, le manque de jeux de données SAR open source, bien annotés ou facilement exploitables, associant texte et image, constitue un obstacle à l’utilisation des modèles fondamentaux existants dans les applications d’observation de la Terre. Dans ce contexte, la génération d’images synthétiques apparaît comme une solution prometteuse pour enrichir ces données rares, permettant ainsi un éventail d’applications plus large. En s’appuyant sur plus de 15 ans d’archives aériennes issues des campagnes d’acquisition de l’ONERA, nous avons constitué un jeu de données d’entraînement complet de 110 000 images SAR, afin d’exploiter un modèle de diffusion latente pré-entraîné de 3,5 milliards de paramètres [1]. Dans ce travail, nous présentons une approche innovante utilisant des techniques de conditionnement spatial au sein d’un modèle fondamental pour transformer des images satellites SAR en représentations SAR aéroportées. De plus, nous démontrons que notre pipeline est efficace pour combler l’écart de réalisme entre les images simulées générées par le simulateur physique EMPRISE de l’ONERA [2]. Notre méthode explore une application clé de l’intelligence artificielle dans l’avancement des technologies d’imagerie SAR. À notre connaissance, nous sommes les premiers à introduire cette approche dans la littérature
Insight from the Kullback--Leibler divergence into aingéndaptive importance sampling schemes for rare event analysis in high dimension
No full textInternational audienceWe study two adaptive importance sampling schemes for estimating the probability of a rare event in the high-dimensional regime d→∞ with d the dimension. The first scheme is the prominent cross-entropy (CE) method, and the second scheme, motivated by recent results, uses as auxiliary distribution a projection of the optimal auxiliary distribution on a lower dimensional subspace. In these schemes, two samples are used: the first one to learn the auxiliary distribution and the second one, drawn according to the learned distribution, to perform the final probability estimation. Contrary to the common belief that the sample size needs to grow exponentially in the dimension to make the estimator consistent and avoid the weight degeneracy phenomenon, we find that a polynomial sample size in the first learning step is enough. We prove this result assuming that the sought probability is bounded away from 0. For CE, insight is provided on the polynomial growth rate which remains implicit. In contrast, we study the second scheme in a simple computational framework assuming that samples from the conditional distribution are available. This makes it possible to show that the sample size only needs to grow like rd with r the effective dimension of the projection, which highlights the potential benefits of these projection methods
Experimental and Numerical Investigations of Magnetohydrodynamic Hypersonic Interactions for High-Altitude Flight
No full textInternational audienceWe present preliminary experimental and numerical results on the interaction between hypersonic ionized plasma flows and magnetized coin-shaped models. In the experiments, a magnetic field is generated by placing cuboid neodymium magnets directly behind models fabricated from various metallic materials. High-speed CMOS camera observations of the plasma-model interaction reveal substantial alterations in plasma self-emission in the presence of the magnetic field, with a pronounced dependence on magnet orientation. These modifications are attributed to high levels of electron magnetization, characterized by Hall parameter values reaching several dozen. Further analysis provides insights into the magnetic field's effects on shock standoff distance and model surface condition. Direct Simulation Monte Carlo (DSMC) numerical simulations are employed to characterize the flow, particularly MHD-relevant quantities such as the full tensorial electrical conductivity. Additionally, a simple "test-particle" solver is used to evaluate charged particle trajectories under the influence of the magnetic field and collisions. These findings offer initial insights into the mechanisms underlying the experimentally observed changes in emission patterns.</div
Configuration of heterogeneous robot fleets for mission planning: A constraint programming-based approach
No full textInternational audienceWith the increasing autonomy of robots and the variety of tasks they are now able to perform, they are increasingly used in complex and critical missions. This paper focuses on the problem of configuring robot fleets for mission planning. In our work, planning a mission, before its execution phase, is based on solving three intrinsically linked configuration problems: the agent configuration problem, denoted ACP, the agent fleet configuration problem, denoted FCP and the mission plan configuration problem, denoted PCP. In this paper, we consider mission planning as a single problem, composed of three interdependent subproblems. A constraint programming (CP) model is proposed for the global problem. Our model is evaluated on the IBM ILOG CP Optimizer solver and on sixty-four artificially generated Search & Rescue instances
A Clustered Approach to Adaptive Kriging Particle Filter for Magnetic Field-Aided Navigation
No full textInternational audienceAutonomous navigation is a key aspect in unmanned mobile robotics, where accurate and efficient localisation and tracking are crucial. One of the challenges in these situations is the reliable use of sensor data, particularly in environments where GNSS is unavailable. Magnetometric navigation provides a promising alternative by leveraging the anomalies of the Earth's magnetic field. However, accurate estimation in such settings is hindered by noises, non-linearities and scarcity of environmental information. Gaussian process regression-based particle filtering can be considered to alleviate these difficulties.In this paper, we propose to improve Gaussian process-based particle filtering techniques by introducing a clustering step to select the relevant training data, improving their computational efficiency and accuracy in the presence of spatial nonstationarities. An auxiliary particle filter is considered due to its robustness to particle degeneracy.The approach was validated on simulated magnetometric data for an unmanned aerial vehicle. Numerical results show that the clustering step greatly reduces the computational cost of the regression while maintaining the accuracy and convergence speed of the particle filter, demonstrating that this technique offers significant advantages for real-time autonomous navigation
Scheimpflug-like processing of differential absorption lidar data on water vapor
No full textInternational audienceDifferential Absorption Lidar (DiAL) method is widely used for a number of ground-based and airborne greenhouse gas measurement applications. However, for a direct detection in the 2µm range, other sources of random noise can be neglected in front of the detection noise which increases with ranging. We therefore propose to evaluate the effect of nonconstant Scheimpflug-like averaging to reduce the signal-to-noise ratio along the line of sight, during a 5-hour measurement campaign in the Paris region
Real-Time Simulation as Operational Validation Means of the European DAA System
No full textInternational audienceThis paper presents the methodology, execution, and results of the Real-Time Simulation (RTS) campaign conducted within the European Detect and Avoid System (EUDAAS) project, focusing on the operational validation of Detect and Avoid (DAA) systems with both Remote Pilots (RP) and Air Traffic Controllers (ATCO) in the loop. The EUDAAS project aims to develop and validate a European DAA solution to safely integrate Remotely Piloted Aircraft Systems (RPAS) into non-segregated airspace. The RTS activities were carried out using a detailed RPAS simulation facility equipped with the DAA model and focused on evaluating both Remain Well Clear and Collision Avoidance functionalities for MALE and Tactical UAV platforms. Human factors were central to the validation process, with emphasis on RP and ATCO perceptions of safety, workload, and system usability. Performance data were collected through system recordings, questionnaires, and interviews. Results confirm the operational feasibility of the EUDAAS DAA concept and provide key insights into user interaction and acceptability in both nominal and non-nominal scenarios. ATCO rated the system’s usability and acceptability positively, due to limited impact on their operational tasks, while RP expressed lower ratings, mainly due some deficiencies that limited full evaluation of the DAA concept. Nevertheless, all participants agreed on the necessity of a DAA system for the effective integration of RPAS into the civil airspace and the findings led to several recommendations that will support the refinement of the future European DAA concept and its system design
Thermal emission in coupled Fabry Perot ultrarefractory metasurfaces
No full textInternational audienceMetasurfaces control over optical properties such as surface emissivity. However, controlling thermal emissivity at high temperatures poses challenges, including issues with melting, thermal expansion mismatches, and oxidation. These factors impose important constraints on the choice of materials and geometric design. In this work, we propose a single-element metasurface made from nanostructured tungsten to mitigate thermal expansion issues, leveraging tungsten's high melting point and thermal stability. This metasurface exhibits a narrow absorption peak (quality factor above 50) which, according to Kirchhoff’s law, directly translates into a thermal emissivity peak. The design employs the coupling of two undercoupled Fabry-Perot-type nano-cavities. Individually, each slit is too large to achieve critical coupling, but through a three-wave interference, this geometry attains critical resonance with an enhanced quality factor compared to single-slit geometries. Additionally, this geometry—with an aspect ratio of 1— is easier to fabricate with clean room processes
Réduction d'ordre de modèles basée sur la collocation sur grilles chimères mobiles pour problèmes paramétriques
No full textInternational audienc
Combustion dynamics of a dual-inlet side-dump ramjet combustor
No full textThis article reports on the results and analyses of an experimental campaign, which aimed at the understanding of the combustion instabilities developing in ramjets. We conducted a series of experiments around an idealized ramjet combustor fed from two symmetric side inlets by air/kerosene streams. The combustor is instrumented, in particular with static and dynamic pressure transducers as well as visualization windows orthogonal to the three main directions of the apparatus. Through which, high-speed recordings of the OH* chemiluminescence are acquired so as to image the combustion dynamics at acquisition rates greater than 23 kHz. Spectral analyses of OH* chemiluminescence recordings using the Spectral Proper Orthogonal Decomposition (SPOD) technique, combined with Fast Fourier Transform (FFT) of dynamic pressure recordings, enable the decomposition of the complex combustion dynamics. The latter develops strong pressure fluctuations. When the combustor outlet is choked, the dominant instability is that of a high-frequency (∼ 3500 Hz), transverse acoustic resonance. When the combustor outlet is unchoked, a lowfrequency (∼ 100 Hz), longitudinal quarter-wave resonance-spanning the full length of the apparatus-dominates the instabilities. The dominant transverse resonance is amplified by one of the aero-acoustic tones induced by the jet-on-jet impingement. This mechanism, referred to as the aero-acoustic-coupling (AAC), is identified for the first time under reacting conditions, and with an inclined jet-on-jet impingement. Another suspected contributor to the intense transverse resonance is fluctuations in the fuel flow rate, induced by upstream-propagating acoustic waves in the inlets, which feeds back into the AAC in the form of heat-release rate fluctuations. Furthermore, we observe the emergence of a ∼ 600-Hz thermo-hydro-acoustic-coupling (THAC), between (i) the full-wave longitudinal resonance of the combustor, (ii) the heat-release, and (iii) a hydrodynamic instability caused by the jet-on-jet impingement. Although this coupling appears to dampen the AAC through non-linear amplifications of neighboring modes of interaction, its intensity remains somewhat sporadic. We present a holistic analysis of how the combustor responds to variations in the Equivalence Ratio (ER) of the air/kerosene mixture. It is observed that the amplitude of the pressure fluctuations is proportional to (i) the thermal power of the engine, and (ii) the coldness of the lateral walls