HAL-CentraleSupelec
Not a member yet
77624 research outputs found
Sort by
Convexity of the annualized cost of energy systems with usage-dependent component lifetimes
No full textAn energy system optimization (joint sizing & dispatch) which accounts for the effect of component aging on the annualized cost of the system is proposed. Despite being nonlinear, the model is proved to be convex under a generalized “lifetime = size/usage” hypothesis. This assumption is shown to cover the classical energy counting model for battery cycling aging, while a linear approximation is needed to account for the operating hours of dispatchable generators. The convexity of the model allows for a piecewise linear approximation which is solved efficiently with linear programming solvers.On an illustrative open-source microgrid sizing test case, the proposed model significantly reduces the levelized cost of energy compared to the classical fixed-lifetime approaches. Solving time is kept within the same order of magnitude, thanks to the model convexity. The proposed model also compares favorably with the more advanced “usage penalty” approaches, even though an improved variant of these is formulated and used
Dissecting pseudoneglect in real and virtual environments: effects of tool and stimulus, not distance or environment
No full textInternational audiencePseudoneglect is a spatial bias favoring the left side of space observed in healthy individuals, typically shown by a leftward deviation when bisecting horizontal lines. Although environmental and individual factors are known to modulate this bias, findings remain inconsistent. This study aimed to provide new insights into the influence of four environmental factors on pseudoneglect: Distance (near, far), Environment (real, immersive Virtual Reality (iVR)), Tool (pen, laser pointer), Stimulus (line, object), while controlling handedness. Fifty-five participants (M = 20.6 years, SD = 2.3, 35 right-handed) completed three line bisection tasks in a real environment (paper-and-pencil at near distance, laser at near and far distances) and three in iVR: (laser at near and far distances, French baguette bisection at near distance). A significant Tool effect emerged: pseudoneglect appeared only in the paper-and-pencil bisection task, while a rightward bias observed with a laser pointer. Stimulus also had a significant effect, with a more rightward bias for the baguette than for lines. No main effects of Distance, Environment, or Handedness were found. Several interpretations are discussed, such as inter-individual variability, coding of peripersonal and extrapersonal spaces, and interaction between semantic and attentional networks. Our results emphasize the need to homogenize methodologies across studies
Virtual reality for gender microaggression training: A qualitative exploration of higher education students’ immersive learning experiences
No full textInternational audienceGender-based microaggressions remain a pervasive challenge for women, compromising their success in academia and the workplace. Despite increased awareness of overt sexism, these subtle forms of bias often go unaddressed, highlighting the need for innovative training approaches that equip individuals with the skills needed to counteract such behaviors. Virtual Reality (VR) has emerged as a promising medium for Diversity, Equity, and Inclusion (DEI) training, offering immersive simulations of real-world scenarios that allow learners to engage in complex social interactions within a computer-controlled environment. This study qualitatively explores higher education students' experiences with VR DEI training through a gender inclusion module focused on navigating microaggressions. Using semi-structured interviews and reflexive thematic analysis, four key themes were developed: (a) immersion and engagement afforded by the VR medium, (b) perceived realism and contextual fidelity of the scenario, (c) perceived learning outcomes, and (d) the practical and technical factors that influenced training delivery. These findings highlight the potential of VR as a pedagogical tool for DEI training, particularly its ability to provide experiential learning opportunities. However, these findings also reveal areas for improvement, including the need for dynamic and adaptive interactions, greater contextual fidelity, and personalized learning tailored to individual learner needs. By addressing these challenges, VR can serve as a practical and impactful training tool for fostering inclusive environments
Identification des patient·es sans-abri ayant reçu des chirurgies en urgence dans l’Entrepôt de Données de Santé de l’Assistance Publique - Hôpitaux de Paris
No full textInternational audienceIntroductionLes parcours chirurgicaux d’urgence des personnes sans-abri sont peu connus. Pourtant, l’expérience du sans-abrisme est associée à des contraintes médicales spécifiques et notamment à un usage moindre des soins primaires en comparaison avec la population générale. Cette étude s’attache à décrire les parcours chirurgicaux non programmés de ces patient·es. La première difficulté réside dans leur repérage, qui passe par l’analyse du texte libre des comptes-rendus (CRs).MéthodesNous sélectionnons dans l’entrepôt de données de santé de l’AP-HP les patient·es ayant subi des chirurgies non programmées entre 2019 et 2022, en utilisant 651 codes CCAM relevant de l’intervention d’urgence. Nous retrouvons tous les documents de leurs dossiers (CR d’hospitalisation, lettre d’orientation, etc.). Nous cherchons à identifier les patient·es sans-abri par la détection de mots-clés dans les documents. L’annotation manuelle des CRs permet d’identifier la Valeur Prédictive Positive (VPP) de chaque mot-clé.RésultatsNous incluons 185 963 patient·es ayant eu des chirurgies urgentes. En sélectionnant sur l’expression « vit/est à la rue », sur 107 CRs sélectionnés au hasard parmi ceux contenant le mot-clé, 103 relèvent de vraies situations de sans-abrisme (VPP : 0.96). Sur le mot-clé « clochard », nous trouvons 115 CRs, dont 13 se réfèrent à de vraies situations de sans-abrisme (VPP : 0.11). Au total, parmi ces 222 CRs annotés manuellement, 116 relevaient effectivement de situations de sans-abrisme. Ces CRs correspondaient à 87 patient·es ayant subi 104 interventions d’urgence (orthopédie : 40, plastique : 21, digestif : 18; cardio-vasculaire : 11; maxillo-faciale : 8; neurologie : 3; urologie : 3).Discussion/ConclusionLes dossiers patient·es informatisés permettent l’identification de patient·es ayant vécu des périodes de sans-abrisme. Néanmoins, la faible prévalence de ces situations requiert des stratégies de repérage spécifiques. Nous nous concentrerons pour la suite sur le développement de méthodes à base de règles (expressions régulières) et d’algorithmes d’apprentissage pour mieux repérer ces patient·es
An Analytical Approach to Near-Field Coupling of Wires With Direct Termination on PCB Ground Plane
No full textInternational audienceThis paper presents an extension to the analytical modeling of electromagnetic near-field coupling, specifically tailored for realistic automotive electronics scenarios. Building upon prior research that modeled coupling between a noise source (metallic loop) and a wire harness in the presence of a printed circuit board (PCB) ground plane, this study introduces a refinement: the direct electrical termination of the wire onto the PCB ground plane. This configuration accurately reflects contemporary EMC testing practices and real-world system integration, where harnesses are often directly connected to electronic boards. The proposed analytical model is developed to account for these altered boundary conditions. Its accuracy and computational efficiency are validated through fullwave simulations using Ansys HFSS and experimental measurements. The results demonstrate good agreement across a broad frequency range, confirming the model's robustness and enhanced realism, while this analytical approach retains its computational speed advantage.</div
RAVE: Rate-Adaptive Visual Encoding for 3D Gaussian Splatting
No full textInternational audienceRecent advances in neural scene representations have transformed immersive multimedia, with 3D Gaussian Splatting (3DGS) enabling real-time photorealistic rendering. Despite its efficiency, 3DGS suffers from large memory requirements and costly training procedures, motivating efforts toward compression. Existing approaches, however, operate at fixed rates, limiting adaptability to varying bandwidth and device constraints. In this work, we propose a flexible compression scheme for 3DGS that supports interpolation at any rate between predefined bounds. Our method is computationally lightweight, requires no retraining for any rate, and preserves rendering quality across a broad range of operating points. Experiments demonstrate that the approach achieves efficient, high-quality compression while offering dynamic rate control, making it suitable for practical deployment in immersive applications. The code is available at https://github.com/inspiros/RAVE.</div
Shared control for assistive and collaborative robotics
No full textInternational audienceAssistive and collaborative robots offer a promising solution for reducing human physical stress and exposure across a broad range of applications. The design and control of such robotic systems should account for humans needs, so as to assist their motion and activity as well as optimize their efficiency and performance throughout the considered tasks. In this respect, designing around the human means accounting for their musculoskeletal constraints, features, and limitations during the development of robotic mechanical systems and control algorithms, such as powered exoskeletons or cobots. This Chapter presents an overview of human-centered design and shared control techniques in robotics, including supporting examples and applications in the field of industrial teleoperation and mobility assistance. We also discuss the perspectives of the field and how such human-centered approaches can be beneficial to a broader range of applications than those where it is implemented nowadays
An enhanced experimental paradigm for auditory BCIs by addressing both acoustic and human factors
No full textInternational audienceThis paper describes an experimental paradigm designed to evaluate the usability of non-sinusoidal amplitudemodulated stimuli for Brain-Computer Interfaces (BCIs) based on steady-state auditory evoked potentials. Instead of the commonly used pure tone, which can be uncomfortable, detectability of alternatives like Brownian noise or natural sounds (e.g. cicada song and cat's purr), which are expected to be more pleasant to listen to, is assessed. Brain responses from 48 participants are recorded while listening to various amplitude-modulated sounds. To ensure accurate sound reproduction, the audio setup's frequency response is measured and compensated for, using an Head And Torso Simulator (HATS). All stimuli are also equalized in loudness to prevent perceived level differences between stimuli. Participants' hearing acuity is measured using an audiometric procedure developed for the needs of the experiment. Participants also completed questionnaires to collect information about their laterality, musicality, age and sex, as these factors may influence auditory perception.</div
Readability as a multi-measure construct in data visualization
No full textInternational audienceIn this paper, we argue that readability cannot be meaningfully discussed without considering multiple complementary measures, and that relying on a single measure constitutes an epistemological choice that constrains the conclusions that can be drawn
Fair effort-sharing for climate change mitigation: A normative approach
No full textInternational audienc