International Professional University of Technology in Nagoya Repository
Not a member yet
15131 research outputs found
Sort by
Qualitative Symbolic Perturbation: Two Applications of a New Geometry-based Perturbation Framework
No full textInternational audienceIn a classical Symbolic Perturbation scheme, degeneracies are handled by substituting some polynomials in ε for the inputs of a predicate. Instead of a single perturbation, we propose to use a sequence of (simpler) perturbations. Moreover, we look at their effects geometrically instead of algebraically; this allows us to tackle cases that were not tractable with the classical algebraic approach
Thermal management for GaN power devices mounted on PCB substrates
No full textThis paper investigates different thermal management solutions for GaN HEMT mounted on Printed Circuit Board (PCB) substrates. Wide bandgap (WBG) power semiconductors like GaN devices have the ability to operate at high switching-frequency (from few 100 kHz to several MHz). To take advantage of their high frequency switching capabilities, the parasitic inductances of power connections as well as the connections between the dies and the gate driver must be minimized. So the majority of GaN chips available on the market are packaged so that they can be directly attached to a PCB. The embedding technology of GaN dies in PCB substrate is attractive because it offers various interconnection possibilities. However, the low thermal conductivity of glass epoxy will result in high thermal resistance of the substrate. So it is of the first importance to seek technological means in order to improve the cooling of GaN chips soldered or embedded in such PCB structures
Large-scale model of flow in heterogeneous and hierarchical porous media
No full textInternational audienceHeterogeneous porous structures are very often encountered in natural environments, bioremediation processes among many others. Reliable models for momentum transport are crucial whenever mass transport or convective heat occurs in these systems. In this work, we derive a large-scale average model for incompressible single-phase flow in heterogeneous and hierarchical soil porous media composed of two distinct porous regions embedding a solid impermeable structure. The model, based on the local mechanical equilibrium assumption between the porous regions, results in a unique momentum transport equation where the global effective permeability naturally depends on the permeabilities at the intermediate mesoscopic scales and therefore includes the complex hierarchical structure of the soil. The associated closure problem is numerically solved for various configurations and properties of the heterogeneous medium. The results clearly show that the effective permeability increases with the volume fraction of the most permeable porous region. It is also shown that the effective permeability is sensitive to the dimensionality spatial arrangement of the porous regions and in particular depends on the contact between the impermeable solid and the two porous regions
SAR array synthesis for next generation earth observation systems
No full textInternational audienceA novel synthesis methodology for fast and robust design of synthetic aperture radar (SAR) arrays for Earth observation is proposed. An innovative integer coding of the discrete problem unknowns (i.e., the magnitude and phase of the array elements in transmission/reception) is introduced to sensibly reduce the dimension of the search space with respect to the standard binary coding and enable the design of large arrangements. Moreover, suitable customizations of the genetic algorithm (GA) operators (i.e., selection, cross-over and mutation) are exploited for an effective exploration of the solutions, by minimizing highly non-linear/unconventional cost functions linked to specific SAR system quality indicators. Some illustrative numerical benchmarks are illustrated in order to verify the effectiveness of the proposed design technique for the synthesis of next generation radars
Driver behaviour in fog is not only a question of degraded visibility - A simulator study
No full textThe aim of this simulator study was to determine whether the effects of fog on driver behaviour were identical for a given road type and whether they could explain fog-related crashes according to road type. Thirty-three participants drove on both two-lane rural roads and motorways according to three visibility conditions (clear weather, 60 m-visibility and 30 m-visibility) and two driving situations (non-free driving and free driving). The variables were: Speeds (Ss), Headway Distances (HDs) and Headway Times (HTs). Fog was simulated using special software designed as part of the French Predit project VOIR, allowing both realistic vehicle headlights and halos to be displayed. The results showed that the drivers decreased their speed with decreasing visibility distance, i.e., speeds were slower in the 30 m-visibility conditions than in clear conditions; but, speeds on the two-lane motorway remained faster than on the two-lane rural road, even for the denser fog. In the 30 m-visibility condition, the faster speeds driven on motorways than on two-lane rural sections violated those advocated by the French Highway Code. The distances travelled in conjunction with the speeds driven according to the two-second rule revealed that HTs less than 2 s and small HDs do not necessarily match with hazardous driving.Le but de cette étude sur le simulateur était de déterminer si les effets du brouillard sur le comportement du conducteur étaient identiques pour un type de route donné et s'ils pouvaient expliquer les collisions liées au brouillard selon le type de route. Trente-trois participants ont conduit sur des routes rurales à deux voies et des autoroutes selon trois conditions de visibilité (temps clair, 60 m de visibilité et 30 m de visibilité) et deux situations de conduite (conduite réglementée et conduite libre). Les variables étaient: Vitesses (Ss), Distances de progression (HD) et temps de progression (HTs). Le brouillard a été simulé à l'aide d'un logiciel spécial conçu dans le cadre du projet PREdit français VOIR, permettant d'afficher les phares et les halos de Îhicules réalistes. Les résultats ont montré que les conducteurs ont diminué leur vitesse avec une distance de visibilité décroissante, c'est-à-dire que les vitesses étaient plus lentes dans les conditions de visibilité de 30 m que dans des conditions claires; mais, les vitesses sur l'autoroute à deux voies sont restées plus rapides que sur la route rurale à deux voies, même pour le brouillard plus dense. Dans la condition de 30 m de visibilité, les vitesses plus rapides sur les autoroutes que sur les zones rurales à deux voies ont violé celles préconisées par le Code de la route. Les distances parcourues en conjonction avec les vitesses conduites selon la règle des deux secondes ont réÎlé que les HT de moins de 2 s et les petites HD ne correspondent pas nécessairement à une conduite dangereuse
A diagonal plus low-rank covariance model for computationally efficient source separation
No full textInternational audienceThis paper presents an accelerated version of positive semidef-inite tensor factorization (PSDTF) for blind source separation. PSDTF works better than nonnegative matrix factoriza-tion (NMF) by dropping the arguable assumption that audio signals can be whitened in the frequency domain by using short-term Fourier transform (STFT). Indeed, this assumption only holds true in an ideal situation where each frame is infinitely long and the target signal is completely stationary in each frame. PSDTF thus deals with full covariance matrices over frequency bins instead of forcing them to be diagonal as in NMF. Although PSDTF significantly outperforms NMF in terms of separation performance, it suffers from a heavy computational cost due to the repeated inversion of big covariance matrices. To solve this problem, we propose an intermediate model based on diagonal plus low-rank covariance matrices and derive the expectation-maximization (EM) algorithm for efficiently updating the parameters of PSDTF. Experimental results showed that our method can dramatically reduce the complexity of PSDTF by several orders of magnitude without a significant decrease in separation performance. Index Terms— Blind source separation, nonnegative matrix factorization, positive semidefinite tensor factorization, low-rank approximation
Silhouette-based Pose Estimation for Deformable Organs Application to Surgical Augmented Reality
No full textInternational audience— In this paper we introduce a method for semi-automatic registration of 3D deformable models using 2D shape outlines (silhouettes) extracted from a monocular camera view. Our framework is based on the combination of a biomechanical model of the organ with a set of projective constraints influencing the deformation of the model. To enforce convergence towards a global minimum for this ill-posed problem we interactively provide a rough (rigid) estimation of the pose. We show that our approach allows for the estimation of the non-rigid 3D pose while relying only on 2D information. The method is evaluated experimentally on a soft silicone gel model of a liver, as well as on real surgical data, providing augmented reality of the liver and the kidney using a monocular laparoscopic camera. Results show that the final elastic registration can be obtained in just a few seconds, thus remaining compatible with clinical constraints. We also evaluate the sensitivity of our approach according to both the initial alignment of the model and the silhouette length and shape
Long-distance WPT unconventional arrays synthesis
No full textInternational audienceTwo innovative array concepts are introduced for the design of long-distance Wireless Power Transfer (WPT) radiating systems. The achievable tradeoffs between complexity/cost mitigation and power focusing capabilities of unconventional WPT architectures with respect to state-of-the-art optimal WPT solutions are investigated. To this end, clustered or sparse WPT arrangements are introduced by formulating their syntheses either as excitation or as pattern matching problems then solved by ad-hoc versions of the Contiguous Partition Method and Compressive Sensing algorithms. Selected numerical examples are presented to assess the features and the potentialities of unconventional WPT designs also in comparison with traditional state-of-the-art optimal methods
Démonstration de MarkPad : Augmentation du pavé tactile pour la sélection de commandes
No full textInternational audienceMarkPad is a technique taking advantage of the touchpad that allows creating a large number of size-dependent gestural shortcuts. It relies on the idea of using visual or tactile marks on the touchpad or a combination of them. Gestures start from a mark on the border and end on another mark anywhere. MarkPad does not conï¿¿ict with standard interactions and provides a novice mode that acts as a rehearsal of the expert mode. We present a working prototype that allows the speciï¿¿cation of spatially organized shortcuts as desired by the user. Mappings of actions with gestures lead to the creation of gestural menus and semantically related groups.MarkPad est une technique prenant avantage du touchpad pour permettre la création d’un grand nombre de gestes dépendants de leur taille. Elle se base sur l’idée d’utiliser des marques visuelles ou visuo-tactiles sur le touchpad ou une combinaison des deux. Les gestes démarrent d’une marque sur le bord et nissent sur une autre n’importe où. MarkPad ne rentre pas en con it avec le pointage et propose un mode novice qui agit comme un mode d’entraînement pour le mode expert. Nous présentons un prototype fonctionnel qui permet de spéci er des raccourcis spatialement organisés selon le souhait de l’utilisateur. Des associations entre des actions et des gestes mènent à la création de menus gestuels, permettant de les regrouper sémantiquement
High-Resolution Semantic Labeling with Convolutional Neural Networks
No full textInternational audienceConvolutional neural networks (CNNs) have received increasing attention over the last few years. They were initially conceived for image categorization, i.e., the problem of assigning a semantic label to an entire input image.In this paper we address the problem of dense semantic labeling, which consists in assigning a semantic label to every pixel in an image. Since this requires a high spatial accuracy to determine where labels are assigned, categorization CNNs, intended to be highly robust to local deformations, are not directly applicable. By adapting categorization networks, many semantic labeling CNNs have been recently proposed. Our first contribution is an in-depth analysis of these architectures. We establish the desired properties of an ideal semantic labeling CNN, and assess how those methods stand with regard to these properties. We observe that even though they provide competitive results, these CNNs often underexploit properties of semantic labeling that could lead to more effective and efficient architectures. Out of these observations, we then derive a CNN framework specifically adapted to the semantic labeling problem. In addition to learning features at different resolutions, it learns how to combine these features. By integrating local and global information in an efficient and flexible manner, it outperforms previous techniques. We evaluate the proposed framework and compare it with state-of-the-art architectures on public benchmarks of high-resolution aerial image labeling