HAL-Ecole des Ponts ParisTech
Not a member yet
42743 research outputs found
Sort by
Improving Multimodal Distillation for 3D Semantic Segmentation under Domain Shift
No full textInternational audienceSemantic segmentation networks trained under full supervision for one type of lidar fail to generalize to unseen lidars without intervention. To reduce the performance gap under domain shifts, a recent trend is to leverage vision foundation models (VFMs) providing robust features across domains. In this work, we conduct an exhaustive study to identify recipes for exploiting VFMs in unsupervised domain adaptation for semantic segmentation of lidar point clouds. Building upon unsupervised image-to-lidar knowledge distillation, our study reveals that: (1) the architecture of the lidar backbone is key to maximize the generalization performance on a target domain; (2) it is possible to pretrain a single backbone once and for all, and use it to address many domain shifts; (3) best results are obtained by keeping the pretrained backbone frozen and training an MLP head for semantic segmentation. The resulting pipeline achieves state-of-the-art results in four widely-recognized and challenging settings. The code will be available at: github.com/valeoai/muddos
A Generalized Potential Game Approach of UAV Swarm Coordination for Hidden Target Localization
No full textInternational audienceConsidering a swarm of Unmanned Aerial Vehicles (UAVs) carrying sensors with nondeterministic detection and noisy localization measurements, while sharing observations with neighboring UAVs, we address the problem of localization of a hidden target in continuous space and discrete time. The goal is to coordinate UAVs to maximize the information gathered while minimizing their individual energy costs. We formulate the problem as a time-varying non-cooperative game with coupling constraints. We show that the target is localized in finite time with probability one and that the game has a generalized potential structure. Further, we provide an exact best-response algorithm for UAVs to iteratively compute their trajectories. Finally, we numerically compare the potential game to the team-based approach, demonstrating comparable performance under different communication graph structures and assessing the impact of the swarm size on various metrics
Modeling Heat and Mass Transfer for a Floating Photovoltaic Power Plant using the Immersed Boundary Method
No full textInternational audienc
CABLE SOLAR : A Tethered Airborne Platform Dedicated To Solar Cell Characterization
No full textInternational audienceA Tethered Airborne Platform Dedicated To Solar Cell Characterization is presented that could be used ultimately as an Aerostat for Solar Power GenerationPV prototype, electronics for data acquisition and intregration on aerostat were tested with success. The first results are promising and could be associated witha simulation part (LAAS collaboration :https://hal.science/hal-04730979
Fast mesoscopic model of plasticity in polycrystals to compute probabilistic S–N curves in high cycle fatigue
No full textInternational audienceHigh cycle fatigue in polycrystals is mostly governed by deterministic laws such as crystal plasticity, but also depends on probabilistic properties, such as random defects and crystallographic and morphological textures, which result in significant scatter of fatigue lifetime at the macroscopic scale. Thus, modeling fatigue phenomena so that the probabilistic density function of failure is anticipated, would be useful especially for very high cycle fatigue involving up to 10 9 cycles. To do so, the grain structure with crystal orientations should be considered in full field computations, which usually involve prohibitive computation cost therefore hindering numerical exploration of statistical distribution of fatigue failures.This paper therefore consists of developing a very fast full field mesoscopic model of polycrystals subjected to crystal plasticity during cyclic loading based on energy minimization techniques. As a result, the uniform plastic slip in each grain is obtained in the form of a relatively simple recursive formula, which guarantees short computation time even for very high cycle fatigue. The proposed approach has been validated against a classical crystal plasticity finite element model in 2D, and satisfying agreement is observed. In addition the model has been applied in combination with classical fatigue criteria to rapidly compute the fatigue lifetime and then derive probabilistic S-N curves, hence creating a substantial link between crystallographic and morphological textures on the one hand, and fatigue lifetime estimations on the other hand.</p
Evaluation of V2X Technologies for the Connectivity of Small Autonomous Vehicles on Secondary Railway Lines
No full textInternational audienceA new concept involving on-demand automatic electric vehicles operating on both road and railway tracks is being developed to rejuvenate secondary railway lines. It is known as FERROMOBILE. Its architecture involves several entities, including vehicles, level crossings, stations, and other infrastructure. The operation of this system necessitates a dynamic, flexible, and safe supervision between these entities, which requires complete and continuous connectivity between them. In this regard, several vehicular communication modes can be created, such as vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I) and vehicle-to-network (V2N). Consequently, connectivity can be achieved by utilizing multiple short and long-range wireless technologies, including vehicle-to-everything (V2X). This paper aims to provide a performance analysis of V2X connectivity technologies for autonomous vehicles on secondary railway lines. It first presents an overview of the potential wireless technologies for FERROMOBILE connectivity. Then, we comprehensively describe the developed emulation platform and perform an experimental evaluation of ITS-G5 and C-V2X for V2V/V2I communications, 4G, 5G NR and LoRa for V2N communications. The evaluation was conducted based on V2X channel models consistent with the FERROMOBILE environment. Several evaluations were performed using latency, jitter, packet delivery rate, and packet error rate as key performance indicators. The evaluation results demonstrate that C-V2X outperforms ITS-G5, and 4G/5G NR surpass LoRa's performance. Indeed, the results show that ITS-G5 and C-V2X can be applicable in V2V/V2I communication and 4G/5G NR in V2N communication for FERROMOBILE architecture
Amélioration de l'Analyse de Cycle de Vie pour les Espaces Verts Urbains : Une Approche Intégrée pour la Prise de Décision
No full textInternational audienceLe développement des espaces verts en milieu urbain suscite de nombreuses attentes : atténuation des îlots de chaleur,amélioration de la qualité de vie voire même réduction des émissions de gaz à effet de serre (Ma et al. 2025). Pourtant,ces bénéfices potentiels et la consommation de ressources nécessaires à leur réalisation (e.g. consommation d’eau)restent encore trop peu intégrés dans les outils d’écoconception appliqués aux projets urbains.Les espaces verts rendent d’importants services écosystémiques (Tardieu et al. 2020). Par exemple, ils participent à laconservation et au développement de la biodiversité, ils réduisent le ruissellement des eaux de pluie ou encore participentà l’atténuation des ilots de chaleur urbains. Pourtant, ces effets ne sont pas pris en compte dans les outils d’écoconceptioncapables de traiter l’échelle quartier. Cet article propose une méthode innovante visant à améliorer la prise en comptedes espaces verts dans les outils d’Analyse de Cycle de Vie (ACV) utilisés à l’échelle quartier. La méthode développéerepose sur des paramètres permettant d’évaluer plusieurs dimensions : la réduction du ruissellement, le stockage decarbone, la biodiversité (spécifique et totale), le confort extérieur, la production de déchets, la consommation d’eau pourl’arrosage, ainsi que la qualité socio-culturelle.Afin de renforcer la précision des modèles existants notamment pour la prise en compte des arbres dans l’ACV des projetsurbains, la méthode proposée intègre des équations allométriques simulant la croissance des arbres. Ceci permetd’estimer précisément leurs impacts environnementaux au cours du temps (e.g. consommation d’eau, entretien, élagage,gestion des déchets d’élagage, etc.).Les données permettant de simuler la croissance des arbres urbains sont encore rares et de qualité hétérogène. Les outilsspécialisés reposent souvent sur des bases de données étrangères (Rötzer et al. 2020). La mobilisation de bases dedonnées françaises permettrait une meilleure prise en compte du contexte spécifique à la France en matière de climat,d’essence et de gestion.14 bases de données d’arbres urbains — contenant des informations telles que la date de plantation, l’âge, la hauteur oule diamètre du tronc — sont aujourd’hui disponibles en open data sur le portail data.gouv.fr. En s’appuyant sur cesbases, un échantillon de 174,886 arbres a été constitué. Cette recherche vient compléter et enrichir les travaux deréférence, notamment ceux de McPherson (2016), fondés sur une base de 15 000 arbres en permettant d’élaborer deséquations allométriques spécifiques au cas français.La méthode proposée est appliquée à un cas d’étude réel, illustrant son potentiel à orienter les décisions d’aménagementdès les phases amont des projets (par exemple en évaluant précisément la quantité de déchets verts récoltés dans lesespaces verts ou les besoins en eau des arbres dans les années suivant leur plantation), ainsi que sa capacité à êtreintégré dans les outils d’ACV à l’échelle du quartier
Incomplete markets with a countable number of states: Equilibrium and no-arbitrage
No full textLa version publiée de l'article est accessible en ligne via son DOI sur JSTOR. The published version of the paper is open online via its DOI on JSTOR.International audienceIn this paper, we prove the existence of an equilibrium in a two-period model à la Hart with incomplete markets and a countable number of states under a mild restriction on the asymptotic behaviour of the returns matrix. Then, we show that an equilibrium asset price is a no-arbitrage price. Conversely, we consider a sequence of equilibria à la Cass corresponding to an increasing number of states associated with a given sequence of present-values. If the sequence of commodity prices has a non-zero cluster point for the product topology, then the limits of these prices and of the allocations (assets, commodities) constitute, together with the given asset price, an equilibrium with a countable number of states
A posteriori error estimates and adaptivity for locally conservative methods. Inexpensive implementation and evaluation, polytopal meshes, iterative linearization and algebraic solvers, and applications to complex porous media flows
No full textInternational audienceA posteriori estimates give bounds on the error between the unknown solution of a partial differential equation and its numerical approximation. We present here the methodology based on H1 -conforming potential and H(div)-conforming equilibrated flux reconstructions, where the error bounds are guaranteed and fully computable. We consider any lowest-order locally conservative method of the finite volume type and treat general polytopal meshes. We start by a pure diffusion problem and first address the discretization error. We then progressively pass to more complicated model problems and also take into account the errors arising in iterative linearization of nonlinear problems and in algebraic resolution of systems of linear algebraic equations. We focus on the ease of implementation and evaluation of the estimates. In particular, the evaluation of our estimates is explicit and inexpensive, since it merely consists in some local matrix-vector multiplications. Here, on each mesh element, the matrices are either directly inherited from the given numerical method, or easily constructed from the element geometry, while the vectors are the algebraic unknowns of the flux and potential approximations on the given element. Our last problem is a real-life unsteady nonlinear coupled degenerate advection-diffusion-reaction system describing a complex multiphase multicomponent flow in porous media. Here, on each step of the time-marching scheme, on each step of the iterative linearization procedure, and on each step of the linear algebraic solver, the estimate gives a guaranteed upper bound on the total intrinsic error, still takes the simple matrix-vector multiplication form, and distinguishes the different error components. It leads to an easy-to-implement and fast-to-run adaptive algorithm with guaranteed overall precision, adaptive stopping criteria for nonlinear and linear solvers, and adaptive space and time mesh refinements and derefinements. Progressively along the theoretical exposition, numerical experiments on academic benchmarks as well as on real-life problems in two and three space dimensions illustrate the performance of the derived methodology. The presentation is largely self-standing, developing all the details and recalling all necessary basic notions
OpenAlex au service de l'évaluation de la recherche : l'idéal d'ouverture à l'épreuve de la qualité des données
No full textInternational audienceOpenAlex est plus qu'une base de données, c'est un projet philosophique et technique ancré dans les valeurs de la science ouverte.3 piliers fondamentaux :- un principe d'ouverture "radicale"- une ambition de couverture exhaustive et inclusive- le potentiel de catalyser une amélioration de l'évaluation de la recherch