Archive ouverte HAL-LAAS
Not a member yet
12189 research outputs found
Sort by
Optimisation des systèmes de contrôle complexes avec des simulateurs différentiables : une approche hybride de l'apprentissage par renforcement et de la planification de trajectoire
No full textInternational audienceDeep reinforcement learning (RL) often relies on simulators as abstract oracles to model interactions within complex environments. While differentiable simulators have recently emerged for multi-body robotic systems, they remain underutilized, despite their potential to provide richer information. This underutilization, coupled with the high computational cost of exploration-exploitation in high-dimensional state spaces, limits the practical application of RL in the real-world. We propose a method that integrates learning with differentiable simulators to enhance the efficiency of exploration-exploitation. Our approach learns value functions, state trajectories, and control policies from locally optimal runs of a model-based trajectory optimizer. The learned value function acts as a proxy to shorten the preview horizon, while approximated state and control policies guide the trajectory optimization. We benchmark our algorithm on three classical control problems and a torque-controlled 7 degree-of-freedom robot manipulator arm, demonstrating faster convergence and a more efficient symbiotic relationship between learning and simulation for end-to-end training of complex, poly-articulated systems.L'apprentissage par renforcement profond (RL) s'appuie souvent sur des simulateurs comme oracles abstraits pour modéliser les interactions au sein d'environnements complexes. Bien que des simulateurs différentiables aient récemment émergé pour les systèmes robotiques multi-corps, ils restent sous-utilisés, malgré leur potentiel à fournir des informations plus riches. Cette sous-utilisation, conjuguée au coût de calcul élevé de l'exploration-exploitation dans des espaces d'état de grande dimension, limite l'application pratique de l'RL en situation réelle. Nous proposons une méthode intégrant l'apprentissage à des simulateurs différentiables afin d'améliorer l'efficacité de l'exploration-exploitation. Notre approche apprend des fonctions de valeur, des trajectoires d'état et des politiques de contrôle à partir d'exécutions localement optimales d'un optimiseur de trajectoire basé sur un modèle. La fonction de valeur apprise agit comme un proxy pour raccourcir l'horizon de prévisualisation, tandis que les politiques d'état et de contrôle approximatives guident l'optimisation de la trajectoire. Nous comparons notre algorithme à trois problèmes de contrôle classiques et à un bras manipulateur robotique à 7 degrés de liberté contrôlé par couple, démontrant une convergence plus rapide et une relation symbiotique plus efficace entre apprentissage et simulation pour l'apprentissage complet de systèmes complexes et polyarticulés
Light emission from a hybrid plasmonic-excitonic STM tunneling junction
No full textInternational audienceThis work focuses on light emission from the nanojunction formed by the tip and a surface in a Scanning Tunneling Microscopy (STM) configuration. The nanojunction includes an ultrathin quantum well made of a single monolayer of a transition metal dichalcogenide material deposited on a gold surface. In this specific configuration, inelastic tunneling of electrons, induced by a bias voltage applied to the tip-surface gap of the STM, excites both Localized Surface Plasmon Polaritons (LSPPs) and excitons. These electromagnetic modes hybridize in this optical nanocavity, producing complex light emission spectra with both plasmonic and excitonic characteristics that depend on the tunneling parameters and the surface roughness. We model the luminescence process as radiative emission triggered by electron tunneling, and we estimate the quantum efficiency as the number of tunneling electrons required to initiate a single exciton recombination and subsequent photon emission. The calculated emission spectra describe the experimental observations well and allow for a thorough understanding of the fundamental physical processes behind light emission in a hybrid plasmonic-excitonic STM nanojunction
Interplay Between Intrinsically Disordered Proteins and Atomically Precise Gold Nanoclusters Modulates their Optical Properties
No full textInternational audienceUnderstanding how structural and optical properties of metallic nanoclusters can be tuned by proteins is crucial for the use of these hybrid molecules in biomedical applications. The interaction of proteins with ultrasmall, atomically-precise gold nanoclusters (Au-NCs) has been mainly investigated in the context of structured proteins, while their behavior with intrinsically disordered proteins (IDPs) remains unexplored. This work examines the structural and optical properties of Au-NCs interacting with bioengineered IDPs containing up to three cysteines. We show that, by exploiting the conformational flexibility of cysteine-containing IDPs, we can anchor proteins to Au-NCs in a position-specific manner, leading to new bioconjugates with properties that differ from those of the individual components. We observed an up to 15-fold photoluminescence enhancement depending on the number of cysteines anchored. By combining mass spectrometry, small-angle X-ray scattering (SAXS), and computational modelling, the ensemble structures of nine bioconjugates with different stoichiometries were elucidated, indicating their overall compactness. Our results suggest that the interface between these atomically-precise species and the conformationally fluctuating protein is responsible for the optical properties of these nanobioconjugates. This research improves our understanding of Au-NC– protein interactions, paving the way to novel nano-molecular hybrid conjugates with tunable properties for bioimaging and therapeutic applications
Quantifying electronic localization induced by defects: implementation of normalized self-interaction in Quantum ESPRESSO
No full textNational audienc
Laboratoire sur puce pour l'isolement de cellules stromales adipeuses circulantes
No full textNational audienceAdipose-derived stem/stromal cells (ASCs) belong to the mesenchymal stromal cell family, renowned for their regenerative potential. Their migration from adipose tissue to other organs has been observed in response to various inflammatory, injury-related, or metabolic stresses. Notably, a high-fat diet has been shown to trigger the release of ASCs from adipose tissue, leading to the formation of ectopic adipocytes (e.g., in skeletal muscle or visceral fat deposits), which are known to contribute to the development of type 2 diabetes. This suggests a potential link between circulating ASC levels and the risk of type 2 diabetes. However, the isolation and characterization of circulating ASCs, which are considered a rare event, remain a significant challenge. Current techniques, including high-sensitivity assays and flow cytometry, have limitations. They struggle to detect the low abundance of ASCs in blood (tens to hundreds of cells per milliliter) or require pre-treatment of blood samples, which can cause cell loss, damage, or irreversible modifications to the molecular state of these rare cells. This thesis aims to develop an innovative device for sorting circulating ASCs, called the "ASC-Finder," based on a two-step negative selection process utilizing two complementary microfluidic modules. The first part of this PhD was dedicated to the development and assessment of the hydrodynamic filtration (HDF) module that was employed for size-based separation of donors' whole blood. Using this label-free approach, we were capable of directly injecting whole blood into our chip with no risk of clogging, without the need for lysis or dilution. The HDF module exhibited excellent erythrocyte depletion rates reaching 99.96 % therefore clearing the major obstacle in ASC detection with no impact on cell integrity. The HDF unit was tested with fluorescent ASCs spiked in blood and yielded total recovery of the original ASCs. The HDF-isolated ASCs were later tested in expansion and differentiation assays and showed identical proliferation and differentiation capacity compared to control groups of ASCs in cell culture media, thereby validating the safety of the device on the fragile primary cells. The second unit of the ASC-Finder device consisted of a leukocyte depletion chip. This module aimed to further purify the ASCs by depleting the remaining leukocytes present in the retentate generated by the first HDF module. We choose to develop a classic Y-Y magnetophoresis chip that deviates leukocytes by the means of antibody coated magnetic beads. Using the ASC-Finder device we were able to obtain high ASC recovery rates from whole blood samples (> 81 %) enabling detection and isolation of these rare-cells ( 81 %), permettant la détection et l'isolement de ces cellules rares (< 100 ASC/mL de sang) sans dommage, ainsi qu'une perte minimale due à la manipulation des échantillons. Le dispositif proposé rivalise avec des dispositifs commerciaux, offrant une étape cruciale vers la caractérisation et l'investigation du rôle des ASCs circulantes dans le sang
Semantic Communications -Necessary for Future Safety-and Mission-Critical Connected Systems
No full textInternational audienceThis article examines how smart technologies are transforming safety-and mission-critical (SMC) systems, such as autonomous vehicles and unmanned aerial vehicles (UAVs). These systems must stay safe even when things go wrong. To do this, they need reliable and real-time communication. However, current methods cannot prioritize important messages based on meaning and typically do not offer more than shutting down the system when communication fails. A new approach called semantic communication (SC) can improve the situation by sending only the meaning of messages. This paper discusses current challenges and how SC can help improve communication and safety in SMC systems
Towards IoT-based Smart Mobility Framework for Proactive Road Stress Detection in Individuals with ASD
No full textInternational audienceAutism Spectrum Disorder (ASD) is a neurodevelopmental disability that significantly increases the difficulties and risks associated with driving. Individuals with ASD often face a variety of challenges, such as increased sensory sensitivities, difficulty adapting to changing environments, and struggles with unexpected situations on the road. These difficulties can lead to sensory overload, panic attacks, and impaired decision-making, all of which increase the risk of accidents and make driving an especially overwhelming task. In this paper, we propose a novel IoT-based smart mobility framework for predictive stress detection in drivers with ASD, enabling the early identification of potential stressors before they encounter them on the road. This approach leverages AI-based models, including LSTM and CNN-based architectures. Unlike existing methods that focus on reactive stress detection, which may be too late, our approach predicts stress triggers in advance, enabling timely and preventive support
Exponentially Stable Stubborn Observers for Discrete-Time Linear Systems (Extended Version)
No full textInternational audienceState estimation is crucial for control and monitoring of dynamical systems, but sporadic disturbances (outliers) can severely degrade the estimator's performance. This paper studies the stability of a so-called "stubborn" observer, designed to mitigate the effects of outliers, for discrete-time linear systems. We prove that the detectability of the system is both necessary and sufficient for global exponential stability of the observer, providing an improvement over existing results that rely on potentially infeasible LMI conditions. The proof is constructive, leading to practical guidelines for selecting the observer parameters. Numerical simulations demonstrate the observer's effectiveness in mitigating outliers and its superior performance compared to a classical Luenberger observer
HackuLoop - An open-science project to create open-source real-time Hardware-In-The-Loop micro-grid systems
No full textNational audienceThis article documents the results of a hackathon focused on developing grid-forming and grid-following inverters. The event highlighted innovation and collaboration in the field of power electronics. This document provides an overview of the objectives, methodologies, and outcomes of the hackathon, along with lessons learned and future directions