Portail HAL des publications du LIRMM
Not a member yet
13279 research outputs found
Sort by
c2c-gem5: Full System Simulation of Cache-Coherent Chip-to-Chip Interconnects
No full textInternational audienceHigh-Performance Computing (HPC) is shifting toward chiplet-based System-on-Chip (SoC) architectures, necessitating advanced simulation tools for design and optimization. In this work, we extend the gem5 simulator to support cachecoherent multi-chip systems by introducing a new chip-to-chip interconnect model within the Ruby framework. Our implementation is adaptable to various coherence protocols, such as Arm CHI. Calibrated with real hardware, our model is evaluated using PARSEC workloads, demonstrating its accuracy in simulating coherent chip-to-chip interactions and its effectiveness in capturing key performance metrics early in the design flow
Finding irrelevant vertices in linear time on bounded-genus graphs
No full textInternational audienceThe irrelevant vertex technique provides a powerful tool for the design of parameterized algorithms for a wide variety of problems on graphs. A common characteristic of these problems, permitting the application of this technique on surface-embedded graphs, is the fact that every graph of large enough treewidth contains a vertex that is irrelevant, in the sense that its removal yields an equivalent instance of the problem. The straightforward application of this technique yields algorithms with running time that is quadratic in the size of the input graph. This running time is due to the fact that it takes linear time to detect one irrelevant vertex and the total number of irrelevant vertices to be detected is linear as well. Using advanced techniques, sub-quadratic algorithms have been designed for particular problems, even in general graphs. However, designing a general framework for linear-time algorithms has been open, even for the bounded-genus case. In this paper we introduce a general framework that enables finding in linear time an entire set of irrelevant vertices whose removal yields a bounded-treewidth graph, provided that the input graph has bounded genus. Our technique consists in decomposing any surface-embedded graph into a tree-structured collection of bounded-treewidth subgraphs where detecting globally irrelevant vertices can be done locally and independently. Our method is applicable to a wide variety of known graph containment or graph modification problems where the irrelevant vertex technique applies. Examples include the (Induced) Minor Folio problem, the (Induced) Disjoint Paths problem, and the F-Minor-Deletion problem
Early detection of Zymoseptoria tritici infection on wheat leaves using hyperspectral imaging data
No full textData collection: The experiment was conducted on two genotypes of durum wheat (Triticum turgidum durum) cultivated under controlled conditions in a growth chamber. Hyperspectral images were collected in planta for twenty-two days to monitor leaf blotch disease caused by the fungal pathogen Zymoseptoria tritici. Images were acquired on the first ligulated leaf in a non-destructive way using two devices: HySpex VNIR-1800 and HySpex SWIR-384 (Norsk Elektro Optikk, Norway).International audienceThis article presents a hyperspectral imaging (HSI) database of healthy leaves and leaves infected with Zymoseptoria tritici fungal pathogen responsible for leaf blotch (Lb) disease. Leaves of two durum wheat genotypes were studied under controlled conditions to track the evolution of Lb disease and capture significant spectral and spatial differences until the onset of symptoms. Hyperspectral image acquisitions were purchased with two cameras in visible-near infrared (VNIR) and short-wave infrared (SWIR) spectral ranges on eighteen dates between one day before inoculation and twenty days after inoculation. For each wavelength range studied, a total of 1175 images provided information on 3326 leaves measured throughout the experiment. These data are valuable since they can be used as a basis to monitor disease's development over time, to build leaf classification models according to their infection status per genotype per day, to develop prediction models related to symptoms' appearance, or to test imaging and spectral analysis methods
XNLP-Completeness for Parameterized Problems on Graphs with a Linear Structure
No full textInternational audienceIn this paper, we showcase the class XNLP as a natural place for many hard problems parameterized by linear width measures. This strengthens existing W [1]-hardness proofs for these problems, since XNLP-hardness implies W [t]-hardness for all t . It also indicates, via a conjecture by Pilipczuk and Wrochna (ACM Trans Comput Theory 9:1–36, 2018), that any XP algorithm for such problems is likely to require XP space. In particular, we show XNLP-completeness for natural problems parameterized by pathwidth, linear clique-width, and linear mim-width. The problems we consider are Independent Set , Dominating Set , Odd Cycle Transversal, (q -)Coloring , Max Cut , Maximum Regular Induced Subgraph , Feedback Vertex Set , Capacitated (Red-Blue) Dominating Set , Capacitated Vertex Cover and Bipartite Bandwidth
RHP Friends, Robot Humanoïde : Combinaison Harmonieuse de Tâches Autonomes et Téléopérées dans un Contexte de Soins Infirmiers
No full textInternational audienceThis paper describes RHP Friends, a social humanoid robot developed to enable assistive robotic deployments in human-coexisting environments. As a use-case application, we present its potential use in nursing by extending its capabilities to operate human devices and tools according to the task and by enabling remote assistance operations. To meet a wide variety of tasks and situations in environments designed by and for humans, we developed a system that seamlessly integrates the slim and lightweight robot and several technologies: locomanipulation, multi-contact motion, teleoperation, and object detection and tracking. We demonstrated the system's usage in a nursing application. The robot efficiently performed the daily task of patient transfer and a non-routine task, represented by a request to operate a circuit breaker. This demonstration, held at the 2023 International Robot Exhibition (IREX), conducted three times a day over three days.Cet article décrit RHP Friends, un robot humanoïde social développé pour permettre le déploiement de robots d'assistance dans des environnements de cohabitation avec des humains. Comme application concrète, nous présentons son utilisation potentielle dans le domaine des soins infirmiers, en étendant ses capacités à manipuler des dispositifs et outils humains en fonction des tâches, tout en permettant des opérations d'assistance à distance. Afin de répondre à une grande variété de tâches et de situations dans des environnements conçus par et pour les humains, nous avons développé un système intégrant de manière fluide ce robot mince et léger avec plusieurs technologies : la locomomanipulation, les mouvements multi-contacts, la téléopération, ainsi que la détection et le suivi d'objets. Nous avons démontré l'utilisation de ce système dans une application de soins infirmiers. Le robot a accompli efficacement une tâche quotidienne de transfert de patient ainsi qu'une tâche non routinière consistant à manipuler un disjoncteur. Cette démonstration a été réalisée lors de l'Exposition Internationale de la Robotique (IREX) 2023, trois fois par jour sur une période de trois jours
Adapting a globally-trained Plant identification model for multi-species detection of invasive alien plants in roadside imagery
No full textConference abstract presented at Living Data 2025. Source Agritrop Cirad (https://agritrop.cirad.fr/616751/) * Autres projets (id;sigle;titre): 101060693;GUARDEN;(EU) safeGUARDing biodivErsity aNd critical ecosystem services across sectors and scales// 101060639;MAMBO;(EU) Modern Approaches to the Monitoring of BiOdiversity//International audienceInvasive alien plants drive biodiversity loss and ecosystem disruption, costing billions globally. Roadsides faciliate spread, yet large-scale monitoring is difficult. Traditional methods like object detection and instance segmentation require costly, time-consuming annotations, impractical for thousands of species. Citizen science platforms like Pl@ntNet use efficient classification models for global identification, yet they struggle with high-resolution multi-species imagery. In this work we adapted the Pl@ntNet global model, trained on millions of images, to detect multiple invasive species in roadside imagery (Espitalier et al. 2025).Pl@ntNet uses a Vision Transformer trained on 6.6 million images covering 43,683 species. Optimized for single-species classification from fixed-resolution (384×384) close-ups of plant organs, it fails on high-resolution complex, multi-scale roadside scenes where plant size varies.We explored two strategies to bridge this domain gap. VaMIS (Variable Model Input Size), forces the model to process the whole roadside scene at once at a larger but non optimal resolution (1024×768). Multi-scale tiling acts like a scanner, decomposing the high-resolution image into hundreds of overlapping patches at multiple zoom levels. Each patch is analyzed independently at the model's native optimal resolution (384×384). If a species is identified with high confidence in any patch, it is considered present in the full image.Both were tested on the Danish Road Dataset (Dyrmann et al. 2021), containing 14,808 high-resolution (4024×3036) vehicle-mounted images for evaluating the detection of six invasive taxa: Cytisus scoparius, Lupinus polyphyllus, Pastinaca sativa, Reynoutria spp., Rosa rugosa, and Solidago spp. 44% of images contain no invasive species, and few include multiple taxa. Performance was measured using balanced accuracy (average recall of presence/absence classes, mitigating imbalance) and AUC (Area Under the ROC Curve, assessing global discrimination of True vs. False Positive rates). F1 and Jaccard index were used on a multi-label test set of 206 images.The tiling approach demonstrated strong generalization, achieving 84.9% balanced accuracy and 91.6% AUC, despite requiring no additional annotations, whereas VaMIS underperformed. To reduce domain shift and re-optimize internal parameters, essential for VaMIS to handle full scenes, we fine-tuned the whole transformer (VaMIS) or the classification head (tiling). Fine-tuning improved both: VaMIS rose from 66.8 to 91.8% accuracy, and tiling from 84.9 to 92.1%. On the multi-species test set, fine-tuned tiling achieved the best results (F1 = 0.85, Jaccard = 0.76), outperforming fine-tuned VaMIS and baseline CNN or YOLO, both models representing former classification and object detection baselines reproduced from previous work (Dyrmann et al. 2021) for comparison.Performance varied by species, but detection of Rosa rugosa consistently exceeded 99% AUC even without fine-tuning. Tiling was about 34x slower but far less resource-intensive to fine-tune (20 min vs 25 h).These results highlight complementary strategies. Tiling offers a generic, training-free pipeline, while VaMIS is faster but needs fine-tuning and memory. Reusing Pl@ntNet models cuts annotation costs, enabling scalable monitoring. Together, these strategies extend biodiversity monitoring to realistic roadside imagery, aligning with FAIR principles
Génération de scénarios de test pour les systèmes de contrôle-commande : une application pour les centrales nucléaires d’EDF
No full textThe work presented in this thesis is part of an industrial project conducted by EDF R&D aiming at automating the generation of tests for the control systems of N4-class nuclear power plants. These systems implement functions described by Logical Diagrams (schematic representations that use AND, OR gates, memories, etc.) executed on Programmable Logic Controllers (PLCs). Throughout the operational lifespan of a nuclear installation, these functions evolve due to regulatory updates (e.g., post-Fukushima), functional improvements related to energy transition requirements, and regular maintenance operations involving component replacements.Ensuring the correct behavior of these systems, preventing regressions in unchanged functionalities, and maintaining compliance with safety standards require rigorous testing procedures. Currently, the generation of test scenarios from logical diagrams is performed manually by expert teams, a process that becomes increasingly complex and costly, particularly when diagrams are large or contain loops between logical blocks.This thesis addresses this challenge by developing formal methods to automatically generate functional test scenarios from Logical Diagrams and functional requirements defined by experts. Given that logical diagrams lack formal semantics, a formal state/transition graph representation has been developed to enable formal graph theory methods for verifying essential properties and generating validation test sequences. However, the exhaustive representation of all possible states and transitions proved to be complex for large-scale diagrams.Consequently, an alternative approach based on Linear Temporal Logic (LTL) and model-checking techniques was proposed. This second method significantly reduces complexity by providing a more concise formal representation of Logical Diagrams. Through this approach, model checking tools can be used to verify important properties on the logical diagrams, and automatically generate functional test scenarios meeting predefined validation objectives. This approach was implemented, tested on real diagrams, and successfully demonstrated its capability to deliver relevant test scenarios within acceptable delays. It has been integrated into a test generation tool aimed at assisting engineers in their test preparations. This work lays the foundation for applying various test generation methods by providing the necessary formal basis. Future enhancements focusing on improving the scalability of the initial graph representation approach are planned to enable structural coverage test generation strategies. Moreover, additional validations involving diagrams of different sizes and structural complexities will be conducted before final integration into the industrial tool developed within the project's framework.Les travaux de cette thèse s’intègrent dans un projet industriel mené par la R&D de EDF visant à automatiser la génération de tests pour les systèmes de contrôle-commande des centrales nucléaires du palier N4. Les fonctions et les commandes remplies par ces systèmes sont exprimées sous la forme de Diagrammes Logiques (schémas comportant des portes ET, OU, des mémoires etc.) implémentés par des Automates Programmables Industriels (API). Les traitements exécutés par les systèmes de contrôle-commande évoluent tout au long de la vie d’une installation nucléaire suite à des évolutions réglementaires (post-Fukushima), suite à des évolutions fonctionnelles (amélioration des performances, impact de la transition énergétique) ou suite à des actions de maintenance (changements de certains composants). L’assurance du bon fonctionnement de ces traitements, la non-régression des parties non modifiées ainsi que le respect des normes de sûreté contrôlés par les autorités concernées fait l’objet de tout une procédure de test. Cette procédure consiste à jouer des scénarios de test sur le système modifié en comparant son comportement aux spécifications. Aujourd’hui, les scénarios de test sont générés manuellement par une équipe d’experts à partir de nombreux diagrammes logiques. Beaucoup de ces diagrammes sont de tailles importantes et comportent des boucles entre les blocs logiques ce qui rend la tâche de génération manuelle assez complexe et coûteuse. Cette thèse s’intéresse à la mise en place de méthodes formelles permettant de générer automatiquement des scenarios de tests fonctionnels à partir des Diagramme Logiques de spécification et de règles métier à valider sur les systèmes de contrôle-commande. Le diagramme logique est un langage métier facilement interprétable par les experts sans qu’il présente de sémantiques formelles. Une définition formelle a été donc établie ainsi qu’une modélisation par des graphes d’états/transitions dans la fin de permettre l’utilisation des méthodes formelles de la théorie des graphes pour la vérification de certaines propriétés intrinsèques ainsi que la génération de séquences de tests de validation. Cependant, la représentation exhaustive de l’ensemble des états et transitions possibles s’est avéré assez complexe pour les diagrammes de grandes tailles. Une deuxième approche de modélisation des diagrammes logiques par les formalismes de la Logique Linéaire Temporelle (LTL) a donc été proposée. Cela a rendu possible l’exploitation des techniques de Model-Checking pour la génération de scénarios répondant à des objectifs fonctionnels de test donnés. Une implémentation de cette approche a été réalisée et testée sur des cas réels de diagrammes et a permis de fournir, dans des délais raisonnables, les scénarios de tests voulus. Plus de tests sur d’autre cas de diagrammes de tailles et complexités structurelles différentes sont prévus avant son intégration dans l’outil final développé dans le cadre du projet
Biosealight: A Low-Cost, Compact, Autonomous, Low-Energy, High-Sensitivity Sensor for Detecting Marine Bioluminescence
No full textInternational audienceBioluminescence, the light emitted naturally by marine organisms, is an important way of communication between organisms in the mesopelagic zone. Nearly 75% of marine organisms, from the surface to the deep sea, and accross a wide taxonomic diversity, use this capability for communication with diverse ecological goals. Bioluminescence detection thus offers an indirect way of tracking the presence of organisms (ranging from zooplankton, dinoflagellates to fishes), their distribution and diel migrations. Such detection can lead, for example, to a better understanding of auto- and heterotrophic plankton communities, vertical migrations of organisms and consequently of a better quantification of the active carbon export in the mesopelagic ocean. However, current technologies still limit large deployments, and high acquisition frequency observations of in situ bioluminescence. This publication aims to describe a new low cost, compact, autonomous, and multi-sensors instrument developed to measure in situ bioluminescence: Biosealight. The sensor's acquisition system is built around a high-sensitivity photon detector that detects photons and converts their number into proportional electrical signals. The photon detection system is integrated with additional sensors that record environmental, including pressure (rated up to 600 bar), temperature, and various system voltages. These measurements are managed by a motherboard featuring an STM32 microcontroller. This motherboard interfaces with a dedicated daughter board, which hosts voltage regulators, the pressure sensor, a high-voltage power supply (>50 V) necessary for the photon detector, amplification electronics, and data storage capabilities consisting of a 1 GB flash memory module with an optional μSD card. Biosealight detects very low light at wavelengths around 450 nm. The acquisition frequency for the bioluminescence sensor is 500 Hz. Other sensors (pressure, temperature) have an acquisition frequency of 38 Hz. High-capacity battery packs power the entire system. The sensor is designed to be easily deployable on a wide range of vectors (such as underwater gliders, CTDs, buoys, trawls...). The sensor being easy to deploy will enable to collect a wide range of bioluminescent data with high spatiotemporal resolution, while recording environmental variables related to the organisms. Various stages of specification research and test protocol under controlled conditions have led to the characterization of the sensor performance in capturing subtle bioluminescent events. It involved calibration procedures designed to verify and optimize key performance parameters such as sensor sensitivity, sampling frequency, and measurable intensity ranges. Dedicated tests were implemented to assess environmental robustness, along with marinization integration protocols, ensuring the sensor maintains operational integrity under challenging marine conditions, including long-term exposure to high pressure cycles and varying temperature. By proposing another way to measure the bioluminescence in situ, and being easy to deploy, this instrument will increase the capacity to observe and charaterize bioluminescence of organisms and improve our understanding of its ecological role
D2.1 - Documentation of available building blocks and technologies
No full textGA nο: 101195295This deliverable illustrates the key building blocks that have been identified as part of the Common European Agriculture Data Space Architecture (CEADS) implementation. Given the extensive work done in other related projects, such as the DSSC and AgriDataSpace, this deliverable is the result of the analysis of potential application domains for dataspaces in agriculture; it introduces the building blocks that will create value while ensuring backward compatibility towards those results already achieved and it also includes specific functionality elements that have been identified as characterising the agricultural sector. While separating the functionality amongst the building blocks, the overall results here shed some light on what incremental integration steps could be to be considered for progressively onboarding functionality and features that make generic DSIs compliant with the expected rules and characteristics of a dataspace. The document is to be considered as a starting point rather than an exhaustive list, and we expect the boundaries and functionality of each of the proposed building blocks to evolve over time, based on requirements elicited from other project activities, especially those related to the implementation of use-cases
Opcode Analysis of Real Encryption-Based Microarchitectural Attacks Using gem5
No full textInternational audienceCan we trust architectural simulators to predict real-world security vulnerabilities? This work utilizes gem5 cycle-accurate simulation to dissect the side-channel leakage of specific AES-NI hardware instructions. By analyzing opcodes like AESENC and AESKEYGENASSIST, we achieved an 89% key recovery rate in simulation, closely mirroring results on physical Intel hardware (96%). Our findings provide a roadmap for identifying microarchitectural footprints and hardening software against Flush+Reload attacks before the code ever touches physical silicon