HAL Arts et Métiers
Not a member yet
14127 research outputs found
Sort by
Optimization of cycloidal propeller performance over multiple operating points using a multi-fidelity CFD-Experimental approach
No full textInternational audienc
Recalage de données BIM en réalité augmentée par une approche de segmentation 3D basée sur la détection 2D multi-vues.
No full textThis thesis presents a hybrid approach to 3D registration in augmented reality, combining object detection in 2D images, segmentation of 3D point clouds, and alignment using the Iterative Closest Point (ICP) algorithm. The proposed method addresses the dual challenge of achieving accuracy and real-time performance, particularly in applications related to Building Information Modeling (BIM).The approach is structured around three main components: (1) automated detection of relevant objects in 2D images using artificial intelligence techniques; (2) segmentation of 3D point clouds through bounding boxes, optimized for mobile device constraints; and (3) registration of the BIM model within the segmented point cloud, initialized through geometric correspondence and refined using ICP.This framework was developed in the context of the R3NDER industrial project, with the aim of ensuring adaptability to other augmented reality scenarios in complex environments. Experimental evaluations focused on specific objects, such as fire hydrants, show that the proposed approach achieves accurate registration with a limited number of views. The use of 3D bounding box intersection offers a relevant balance between spatial accuracy and algorithmic complexity.Cette thèse présente une approche hybride pour le recalage 3D en réalité augmentée, combinant détection d’objets dans des images 2D, segmentation de nuages de points 3D, et recalage par l’algorithme Iterative Closest Point (ICP). L’objectif est de répondre aux exigences de précision et de performance en temps réel posées par les applications de réalité augmentée dans le cadre du Building Information Modeling (BIM).L’approche proposée s’organise en trois modules complémentaires : (1) une détection automatique d’objets d’intérêt dans des images 2D par des méthodes d’intelligence artificielle ; (2) une segmentation des nuages de points 3D à l’aide de boîtes englobantes, adaptée aux contraintes de calcul des dispositifs mobiles ; (3) un recalage du modèle BIM dans le nuage de points segmenté, en s’appuyant sur une initialisation par correspondance géométrique suivie d’un raffinement par ICP.Cette approche a été développée dans le contexte du projet industriel R3NDER, tout en visant une portabilité vers d’autres cas d’usage en réalité augmentée dans des environnements complexes. Les expérimentations menées sur des objets tels que les bornes d’incendie montrent que l’approche permet un recalage précis avec un nombre limité de vues. L’intersection des boîtes englobantes s’est révélée pertinente pour équilibrer la précision spatiale et la complexité algorithmique
Elasticipy: A Python package for linear elasticity and tensor analysis
No full textInternational audienceElasticipy is a Python library designed to streamline computation and manipulation of elasticity tensors for materials and crystalline materials, taking their specific symmetries into account. It provides tools to manipulate, visualise, and analyse tensors –such as stress, strain and stiffness tensors– simplifying workflows for materials scientists and engineers
Multimodal measurement of the mental workload during an assembly and disassembly task
No full textInternational audienceMental workload overload is a major cause of human error in industrial tasks such as maintenance. Human errors can compromise not only system safety but also lead to high social and economic costs, reduce equipment productivity, and cause incidents, accidents, and fatalities. To this day, we do not have an adequate assessment of mental workload in maintenance, which would help design maintenance processes more effectively by incorporating this crucial aspect. The objective of this study is to determine the ability of our indicators to measure mental workload during a disassembly and assembly task in a laboratory condition. Thirty-six participants performed a disassembly and assembly task under two different mental workload conditions. Subjective measures (NASA-TLX), performance metrics (number of errors), and cardiovascular data (heart rate, heart rate variability, and breathing rate) were analyzed. We observed a higher number of errors and elevated NASA-TLX scores in the high mental workload condition. Regarding cardiovascular data, interesting trends in the temporal domain were observed despite mostly non-significant results. Although conducted in a laboratory, this multimodal mental workload measurement method is promising for diagnosing and understanding operators' cognitive behavior, and deserves validation in real-world maintenance conditions
Topology Optimization of Chip Inductor Using Density Method
No full textInternational audienceThis paper proposes a novel methodology of the topology optimization method considering eddy current effects. The method is applied on chip inductors modelled by the Finite Element Method (FEM). Aiming to meet a specified inductance value while minimizing eddy current losses, we employ a density-based approach to construct a continuous material distribution. The derivative of the objective function with respect to the material distribution is obtained using the adjoint variable method, then the material layout is iteratively updated via the L-BFGS-B algorithm. The proposed framework is validated on both single-turn and multi-turn inductor structures, achieving designs that satisfy the target performance within a limited number of iterations. A key innovation of this work lies in the integration of field-circuit coupling into the topology optimization framework, enabling the analysis of inductors under complex coil configurations involving both series and parallel connections. Additionally, we present an original derivation of the sensitivity formulation associated with the inductance value ensuring that the optimized inductance meets the design specification
Optimisation and performance comparison of direct and parametric piezoelectric energy harvester with geometrical nonlinearities
No full textInternational audienceThe behaviour and performance of a flexible beam piezoelectric energy harvester (PEH) in direct and parametric excitation, including the effect of geometrical nonlinearities, are addressed in this paper. First, the electromechanical modelling of the harvester is addressed. A Timoshenko geometrically exact model of a laminated piezoelectric beam in large rotation, including parametric excitation, is proposed, extending previous results of the literature. Then, considering cantilever boundary conditions, it is simplified under Euler-Bernoulli, inextensible assumptions and third order Taylor expansion, suitable for modal expansion. Two perturbation methods are tested to compute vibratory response under direct and parametric excitations. They are both found inaccurate for large amplitude oscillations, leading to a preference for numerical solving by continuation of periodic solutions. PEH behaviours and performances are finally carefully estimated regarding the optimal harvested power in a shunted resistor at resonance. Comparisons between direct and parametric forcing are proposed and the effect of geometrical nonlinearities is estimated. An interesting result is that the harvested power under parametric excitation cannot compete with the one under direct excitation.</div
III-Nitride MEMS drum resonators on flexible metal substrates
No full textInternational audienceWe present a simple and efficient process for fabricating III-Nitride (III-N) mechanical resonators on flexible metal substrates. This method combines Van der Waals epitaxy of III-N epilayers with the deposition of a thick metal stressor atop the III-N layers. During thermal treatment, the 30 μm thick metal stressor deposited on a 300 nm AlGaN/500 nm GaN layer grown on a 3 nm two-dimensional hexagonal-Boron Nitride (2D h-BN) release layer, initiates a one-step Self-Lift-Off and Transfer (SLOT) process. This process effectively transfers the III-N heterostructure from the h-BN/Sapphire growth wafer to the flexible metal stressor substrate. Additional local etching of the metal stressor and deposition of front electrodes allow for releasing self-standing III-N layers with integrated actuation. Fabricated III-N MEMS drum resonators were analyzed using optical profilometry and laser Doppler vibrometer, enabling the observation of static deflections and distinct vibration modes. Finite element method (FEM) simulations were also performed to further understand experimental observations and assess the mechanical properties of the released III-N layers, particularly enabling the estimation of stress in the GaN and AlGaN released layers. This straightforward approach not only provides a practical solution for cost-effective III-N MEMS resonators but also ensures flexibility, and crack-free structures
High-fidelity Haptic Surface: Development of Controlled Friction Interfaces
No full textHaptique et vibrotactile : approche physique et transducteurs; GTEA - Transducteurs et Électroacoustique: GSAM - Acoustique MusicaleNational audienceHaptics diffused in our everyday devices from the smartphone and the vibrotactile alert to other sectors like healthcare and automotive applications. Recent advances have seen high-fidelity haptics emerging that can enhance communication from human to the machine. The friction based tactile interfaces recreate the modulation of friction forces that exist when touching a rough surface. Texture can then used as a language in Human Computer Interfaces. Using a vibrating plate at ultrasonic frequency, friction can be reduced with a fingertip. If the vibration amplitude can be modulated, high-fidelity haptics can be achieved. The paper presents the design principles of a high fidelity haptic interface which exploits a transversal mode of a glass plate. A closed loop control, which helps to compensate for damping which occurs whe the finger touches the interface is developped. Then, multimodal excitation of the plate is presented in order to confine the vibration to a specific location on the plate
Observatoires Milieux-Sociétés et Données Géographiques: Enjeux Méthodologiques et Perspectives
No full textCet article présente un panorama méthodologique des observatoires scientifiques qui considèrent les dimensions milieux et sociétés de phénomènes sur des territoires d'étude et s'intègrent dans les systèmes de décisions territoriales. Il aborde et étudie les écosystèmes, les motivations, les cadres théoriques, les méthodologies et technologies mises en œuvre par ces observatoires orientés vers la modélisation des phénomènes spatio-temporels utiles à la décision. Nous étudions ces observatoires à partir des principes de représentation du temps et de l'espace dans les processus d'acquisition et de partage des données, leur contextualisation au regard de différentes modalités d'utilisation jusqu'aux interactions espace-temps société et leur mise en œuvre dans des applications environnementales et urbaines. Les progrès actuels de la recherche dans la mise en œuvre de ces observatoires et les problématiques encore ouvertes seront discutés
