HAL Portal UTC Université de Technologie de Compiègne
Not a member yet
11652 research outputs found
Sort by
Optimizing Algorithm for Existing Fiber-Optic Displacement Sensor Performance
No full textInternational audienceThis paper describes the optimal design of a miniature fiber-optic linear displacement sensor. It is characterized by its ability to measure displacements along a millimetric range with sub-micrometric resolution. The sensor consists of a triangular reflective grating and two fiber-optic probes. The measurement principle of the sensor is presented. The design of the sensor's triangular grating has been geometrically optimized by considering the step angle of the grating to enhance the sensor's resolution. The optimization method revealed a global optimum at which the highest resolution is obtained
Cartographie interactive des acteurs de l'écosystème règlementaire des dispositifs médicaux
No full textSeveral skateholders are involved at each stage of the life cycle of a Medical Device. Although they all have different activities and obligations, they still have certain characteristics in common. Indeed, they all are interdependent on one another to ensure the quality, safety and performance of the medical devices made available to patients. Above all, they all are influenced by the regulatory and standards framework in which they operate. This ensemble (regulations and players) is known as the medical device regulatory ecosystem. The dynamics of this ecosystem took a radical turn when the new regulation on Medical Devices (Regulation (UE) 2017/745 and Regulation (UE) 2017/746) came into force on the 26th May 2021. Since then, the ecosystem may appear to be very extensive, complexly structured and too fast-moving for a novice professional. And yet, if they want their device to be marketed and used, they will inevitably have to interact with the other members of their sector. A potential solution has therefore emerged : the creation of an interactive tool summarizing the composition of this sector.Plusieurs acteurs interviennent à chacune des étapes du cycle de vie d'un dispositif médical. Bien qu'ayant tous des activités et des obligations différentes, des caractéristiques communes peuvent tout de même leur être trouvées. Tous agissent, en effet, en interdépendance les uns des autres afin d'assurer la qualité, la sécurité et les performances des dispositifs médicaux mis à disposition des patients. Et surtout : tous sont influencés par le cadre réglementaire et normatif dans lequel ils s'inscrivent. Cet ensemble (règlementation, normes et acteurs) est appelé l'écosystème règlementaire des dispositifs médicaux. Les dynamiques de cet écosystème ont connu un tournant radical lorsque le 26 mai 2021, la nouvelle règlementation autour des dispositifs médicaux (Règlement (UE) 2017/745 et Règlement (UE) 2017/746) est entrée en application. Depuis, l'écosystème peut paraître comme étant très étendu, complexement structuré et trop évolutif pour un professionnel néophyte. Et pourtant, s'il souhaite que son dispositif soit commercialisé et utilisé, il devra forcément interagir avec les autres membres de sa filière. De là émerge donc une potentielle solution : la création d'un outil interactif synthétisant la composition de ce secteur
More powerful energetic reasoning using redundant resources for the Cumulative Scheduling Problem
No full textInternational audienc
Superelastic and biocompatible Ti-Zr-Nb-Sn coatings deposited by magnetron sputtering
No full textInternational audienceWe report on the microstructural, mechanical and biological properties of novel quaternary TiZrNbSn (in at.%) coatings obtained by magnetron sputter-deposition on Si substrates and Ti disks. In particular, the influence of Nb content on the phase composition and superelastic response of as-deposited Ti-22Zr-xNb-3Sn films are examined. Three compositions were selected corresponding to Nb content x = 10, 17 and 20 at.%. The structure of the films containing 10 and 17 at.% Nb consists of a mix of alpha'' and (3 phases in different proportions while the film with higher Nb content (20 at.%) possesses a single (3 structure. Their superelasticity was studied using a nanoindenter equipped with a spherical tip of 50 mu m diameter. It revealed that the film containing 17 at.% Nb exhibited the highest superelastic behavior with the largest indentation depth recovery ratio values all over the broad range of maximum penetration depths investigated as compared to the values measured for the two other films with 10 and 20 at.% Nb. The biocompatibility tests showed that Ti-22Zr-xNb-3Sn coatings involve high cytocompatibility and do not induce haemolysis, similarly to reference materials (Ti and NiTi)
Discrete Minimax Binary Relevance Classifier for Imbalanced Multi-label Classification
No full textInternational audienceMulti-label classification (MLC) is a supervised learning problem where each instance can be associated with none, one, or multiple labels. MLC has received increasing attention due to its wide range of applications, such as text categorization and medical diagnosis. Despite a rich literature on MLC, handling imbalanced data, often encountered in real-world MLC datasets, has not been tackled satisfactorily. Based on a thorough literature review, it appears that the existing methods for imbalanced MLC are either hard to be coupled with sound theoretical guarantees or of limited scalability. This paper discusses the potential (dis)advantages of existing methods for imbalanced MLC, when being coupled with Binary relevance classifier (BRC), and introduces Discrete Minimax BRC (DMBRC), which would be a promising attempt to robustify the BRC by leveraging theoretically sound properties of the Discrete Minimax Classifier. We also provide empirical evidence to illustrate how DMBRC may be advantageous in balancing the label-wise error rates. Finally, we envision future works on further strengthening DMBRC in both label-wise error rates and conventional MLC evaluation metrics
Qualification des Modalités du Toucher Social dans les Interactions Humain-Agent en Réalité Virtuelle
No full textInternational audienceLes interactions sociales et la communication impliquent de multiples canaux sensoriels. Parmi ces canaux, le toucher peut être utilisé pour transmettre des émotions et des intentions communicatives, et contribue à la création et au renforcement des liens sociaux. Ajouter le toucher social aux capacités sociales des agents conversationnels animés (ACA) en réalité virtuelle (RV) pourrait améliorer leurs capacités communicatives. Mais le toucher social est lui-même multimodal, car il implique souvent un geste visuel, une sensation haptique, voire le son du contact peau à peau. Dans le cadre de ce projet en cours, nous présentons un protocole d’étude dans lequel nous explorons les contributions respectives des modalités visuelle, audio et tactile à la sensation d’être touché par un geste de toucher social effectué par un ACA en RV. Nous évaluons également à quel point un tel toucher est perçu comme étant réellement provoqué et réalisé par l’ACA selon les différentes modalités sensorielles impliquées et leurs modulations
Multi-Risk Assessment and Management in the Presence of Personal Light Electric Vehicles
No full textInternational audienceThis paper presents an approach to autonomous vehicle navigation in urban environments with dynamic and multi-modal agents like Personal Light Electric Vehicles (PLEVs). The traditional Predictive Inter-Distance Profile (PIDP) risk assessment metric (Bellingard et al., 2023) is extended to handle multiple multi-modal motions using a fusion of PIDPs (F-PIDP). This approach accounts for the uncertainties in the various trajectories that PLEVs can follow on the road. A priority-based strategy is then developed to select the most dangerous agent. Then F-PIDP and Model Predictive Control (MPC) algorithm is employed for risk management, ensuring safe and reliable navigation. The efficiency of the proposed method is validated through several simulations
Robust Discrete Bayesian Classifier Under Covariate and Label Noise
No full textInternational audienceIn this paper, we focus on the Discrete Bayesian Classifier (DBC), which discretizes the input space into regions where class probabilities are estimated. We investigate fuzzy partitioning as an alternative to the hard partitioning classically used to discretize the space. We show that our approach not only boosts the DBC’s performance and resilience to noise, but also mitigates the loss of information due to discretization. The benefits of soft partitioning are demonstrated experimentally on several synthetic and real datasets
Synthesis and characterization of core–shell magnetic molecularly imprinted polymer nanocomposites for the detection of interleukin-6
No full textInternational audienceInterleukin-6 (IL-6) belongs to the cytokine family and plays a vital role in regulating immune response, bone maintenance, body temperature adjustment, and cell growth. The overexpression of IL-6 can indicate various health complications, such as anastomotic leakage, cancer, and chronic diseases. Therefore, the availability of highly sensitive and specific biosensing platforms for IL-6 detection is critical. In this study, for the first time, epitope-mediated IL-6-specific magnetic molecularly imprinted core-shell structures with fluorescent properties were synthesized using a three-step protocol, namely, magnetic nanoparticle functionalization, polymerization, and template removal following thorough optimization studies. The magnetic molecularly imprinted polymers (MMIPs) were characterized using dynamic and electrophoretic light scattering (DLS and ELS), revealing a hydrodynamic size of 169.9 nm and zeta potential of +17.1 mV, while Fourier transform infrared (FTIR) spectroscopy and fluorescence spectroscopy techniques showed characteristic peaks of the polymer and fluorescent tag, respectively. Scanning electron microscopy (SEM) and high-resolution transmission electron microscopy (HRTEM) investigations confirmed the successful encapsulation of the magnetic core within the ca. 5-nm-thick polymeric shell. The MMIP-based electrochemical sensing platform achieved a limit of detection of 0.38 pM within a linear detection range of 0.38-380 pM, indicating high affinity (dissociation constant K D = 1.6 pM) for IL-6 protein in 50% diluted serum samples. Moreover, comparative investigations with the non-imprinted control polymer demonstrated an imprinting factor of 4, confirming high selectivity. With multifunctional features, including fluorescence, magnetic properties, and target responsiveness, the synthesized MMIPs hold significant potential for application in various sensor techniques as well as imaging
Détection continue du stress à partir de signaux physiologiques
No full textThe primary aim of this thesis is to establish a reliable indicator of stress presence based on physiological signals, a critical area given the complexity of stress and its widespread impact on mental and physical wellbeing. The approach taken is geared towards real-time monitoring of stress in ambulatory contexts, thereby empowering individuals to actively manage their health in everyday settings outside traditional clinical environments. To achieve this goal, the research is structured around two main objectives. Firstly, it involves collecting a comprehensive and consistent stress database. This initiative aims to address inconsistencies in data collection methods and participant demographics seen in existing datasets, providing a unified and robust resource for future research. Secondly, the thesis focuses on developing stress detection models. Using advanced data science techniques, the objective is to create models capable of accurately and swiftly identifying stress indicators from collected physiological signals. This approach includes extracting diverse features from each physiological signal and training Al models using both unimodal and multimodal approaches. In parallel, significant emphasis is placed on the generalizability of stress detection models. This aspect is crucial to ensuring that the models remain relevant, accurate, and applicable across various demographic and environmental contexts. Divided into several chapters, this thesis also explores the physiological processes influenced by stress, relevant algorithmic strategies, and the potential implications of its findings for stress management and prevention in the near future.Le but principal de cette thèse est d'établir un indicateur fiable de la présence de stress à partir de signaux physiologiques, un domaine crucial étant donné la complexité du stress et ses effets étendus sur le bien-être mental et physique. L’approche adoptée vise à permettre une surveillance en temps réel du stress dans des contextes ambulatoires, offrant ainsi aux individus la possibilité de gérer activement leur santé au quotidien en dehors des environnements cliniques traditionnels. Pour atteindre cet objectif, la recherche est structurée autour de deux objectifs principaux. Tout d'abord, elle s’engage dans la collecte d’une base de données exhaustive et cohérente sur le stress. Cette initiative vise à surmonter les incohérences observées dans les méthodes de collecte de données et les caractéristiques démographiques des participants des bases de données existantes, afin de fournir une ressource unifiée et robuste pour la recherche future. Ensuite, la thèse se concentre sur le développement de modèles de détection du stress. En utilisant des techniques avancées de science des données, I'objectif est de créer des modèles capables d'identifier de manière précise et rapide les indicateurs de stress à partir des signaux physiologiques collectés. Cette approche inclut l’extraction de caractéristiques diverses à partir de chaque signal physiologique et l’entraînement de modèles d'intelligence artificielle en utilisant des approches tant unimodales que multimodales. Parallèlement à ces efforts, un accent particulier est mis sur la généralisabilité des modèles de détection du stress. Cette dimension est cruciale pour assurer que les modèles restent pertinents, précis et applicables dans une variété de contextes démographiques et environnementaux. Cette thèse, divisée en plusieurs chapitres, explore également les processus physiologiques influencés par le stress, les stratégies algorithmiques pertinentes et les implications potentielles de ses résultats pour la gestion et la prévention du stress dans le futur proche