Archive ouverte HAL-LAAS
Not a member yet
12189 research outputs found
Sort by
Optimizing Energy in Supervised Learning with Data Summarization: A Comparative Study
No full textBonjour, C'est la version preprint que j'ai soumise initialement, donc j'ai le droit de la publier.International audienceThe signicant computational demands of modern machine learning (ML) models raise growing concerns about their environmental impact and economic cost during training. To address this, data summarization techniques, which involve selecting a smaller, representative subset of the training data, oer a promising avenue for optimizing energy consumption without compromising model performance. This article investigates the inuence of various data summarization algorithmsspecically random sampling, Facility Location (FL), and CRAIGon three critical aspects of supervised learning (SL) training: model accuracy, energy consumption, and overall eciency, which we quantify as the ratio of accuracy to energy consumption. Our extensive experimental ndings reveal that while sophisticated methods like FL and CRAIG aim to select highly representative subsets, random sampling consistently achieved a robust balance of accuracy and eciency. This is particularly evident when accounting for the often signicant pre-processing energy overheads incurred by more complex selection strategies. Furthermore, we emphasize the pivotal role of early stopping criteria in optimizing the overall energy eciency of the training process. Our analysis demonstrates that strategic adjustments to these criteria can substantially reduce the number of training epochs required for convergence, thereby mitigating considerable energy waste
Measurement of the refractive index modulation induced by Quantum Confined Stark Effect for active metasurfaces realisation.
No full textInternational audienceAchieving an active optical index modulation in materials is an enabling asset for the realisation of active metasurfaces.Ideally the modulation should be large and ultra-fast and without adding losses to the system.Multiple simulation works already showcase how high reflectivity and fine phase tuning could be obtained with semiconductor nano-resonators [1,2].It is thus interesting to concentrate our efforts of refractive index modulation for this kind of materials.The Quantum Confined Stark Effect (QCSE) is an ultrafast electro-optic effect which has been studied for some times already [3].It is known to induce changes in the absorption of a structure near the gap by tuning transition energies.At wavelengths larger than the bandgap, it is possible to have an ideally null absorption but still achieve a finite refractive index modulation.Inserting those active components in highly resonant structures, one can fabricates highly reflective active metasurfaces with ultra-fast tunability.An active metasurface exploiting such an electrooptic index modulation has been published in 2019 by Atwater teams [4].In this work, a PIN-type structure was used to apply an electric field in the quantum wells. An index modulation of 0.01 was expected but not directly measured.In this study, we present the design of a NIN-type, GaAs/AlGaAs structure and its refractive index electro-modulation at wavelengths below the band-gap.Ellipsometric measurements of the structure are used to precisely determine the different optical properties of the structure.These parameters are used to model the reflectivity modulation that can be induced by a refractive index modulation.These predictions are then compared to the experimentally measured reflectivity modulation induced by an external electric bias.A refractive index variation as high as 0.02 was deduced from measurements at wavelengths in the QW transparency range. This value is in good agreement with theoretical predictions of the Stark effect (Nextnano simulations).This refractive index variation measurement in electroreflection is an important milestone for the conception and development ofactive full dielectric metasurfaces exploiting the intrinsic speed of QCSE optical index modulation.[1]: M. Elsawy et al., Laser & Photonics Review, 2023, 10.1002/lpor.202200880[2]: B. Yang et al., Optics Express, 2025, 10.1364/OE.563702[3]: Y. Kan et al., IEEE Journal of Quantum Electronics, 1987, 10.1109/jqe.1987.1073288[4]: P. C. Wu et al., Nature Com., 2019, 10.1038/s41467-019-11598-
Control of Humanoid Robots with Parallel Mechanisms using Differential Actuation Models
No full textSeveral recently released humanoid robots, inspired by the mechanical design of Cassie, employ actuator configurations in which the motors are displaced from the joints to reduce leg inertia. While studies accounting for the full kinematic complexity have demonstrated the benefits of thesedesigns, the associated loop-closure constraints greatly increase computational cost and limit their use in control and learning.As a result, the non-linear transmission is often approximated by a constant reduction ratio, preventing exploitation of the mechanism’s full capabilities. This paper introduces a compact analytical formulation for the two standard knee and ankle mechanisms that captures the exact non-linear transmission while remaining computationally efficient. The model is fully differentiable up to second order with a minimal formulation,enabling low-cost evaluation of dynamic derivatives for trajectory optimization and of the apparent transmission impedance for reinforcement learning. We integrate this formulation into trajectory optimization and locomotion policy learning, and compare it against simplified constant-ratio approaches. Hardware experiments demonstrate improved accuracy and robustness, showing that the proposed method provides a practical means to incorporate parallel actuation into modern control algorithms
Hierarchical Modelling of the Impact of Obsolescence on System Availability
No full textInternational audienceTechnological advances, changing needs and market dynamics generate a multitude of obsolescence issues in almost all sectors of activity. This represents serious challenges for companies, particularly in terms of quality, availability and maintainability. These challenges are particularly acute for complex systems with a long lifespan such as trains or planes, which must maintain acceptable levels of performance for many years. The obsolescence of components, documentation, tools and personnel skills are all industrial risks that must be controlled. In this context, this research presents models to represent the impact of obsolescence on the availability of complex systems. The objective is to show the mechanisms of these impacts. These models, built using Petri nets, then make it possible to estimate the possible degradation of availability following the occurrence of obsolescence, an aspect not addressed in this article. The modelling is based on a principle of classifying components into four classes, allowing to model the multilevel architecture of a system. By associating a set of Petri models with each class, and by synchronising the models between them, the impact of obsolescence on availability is clearly described. The article ends with a number of conclusions and in particular with a presentation of the research carried out to predict the availability of systems in the presence of obsolescence of components, documentation, personnel or tools
Safe and wind-aware synchronous path planning for a fleet of fixed-wing constant speed aircraft
No full textPath planning for multiple vehicles is a difficult task, but even more so for fixed-wing aircraft. When flying several of them, multiple constraints may apply: constant airspeed, to maintain optimal operational conditions; synchronous arrival to destination, to achieve formation flight; taking into account wind, for limiting deviation from plan, especially for smaller Unmanned Aerial Vehicles. We present a method to solve this problem based on enumerating variations of Dubins paths until a conflict-free solution is found. This provides a simple and parallelizable scheme that can be extended to handle sequencing tasks. We showcase different situations solved in simulation to illustrate the possible applications: transitions between formations, getting into formation and airport arrival sequencing
Design and Building of a Cost-Effective Six-Component Optical Borehole Strainmeter
No full textInternational audienceDue to their high resolution and near real-time capabilities, borehole strainmeters are a critical complement of satellite-based geodetic systems [GPS/global navigation satellite system (GNSS)] and satellite interferometry. However, commercial strainmeters remain expensive and only provide the three horizontal components of the strain tensor. We propose a novel design able to detect the six components of the 3-D strain tensor at relatively low cost. Here, we embed six compliant elastic gauges in a sphere in order to evenly sample space directions. Each gauge exhibits an amplification ratio of ~30. Interrogated by Fabry-Perot interferometers illuminated by a single laser diode (LD) through a multichannel fiber cable, these opto-mechanical systems exhibit a resolution of ~89 pm/ √ Hz over a dc – 500-Hz bandwidth, ultimately providing a strainmeter resolution <1 nanostrain. A supplementary interferometric device is also implemented to detect and correct nongeometrical optical phase changes due to pressure and temperature variations. An original building method allowed us to mold fibered cement around a thin sphere equipped by the optical strain gauges. Moreover, an adjustable pressure device is integrated to allow in situ calibration in order to correct the 3-D strain tensor from borehole and cement heterogeneities. This cost-effective strainmeter was successfully installed in November 2023 in a 30-m deep borehole at the Larzac Observatory in the French Massif Central
Optimal Control of Walkers with Parallel Actuation
No full textInternational audienceLegged robots with closed-loop kinematic chains are increasingly prevalent due to their increased mobility and efficiency. Yet, most motion generation methods rely on serial-chain approximations, sidestepping their specific constraints and dynamics. This leads to suboptimal motions and limits the adaptability of these methods to diverse kinematic structures. We propose a comprehensive motion generation method that explicitly incorporates closed-loop kinematics and their associated constraints in an optimal control problem, integrating kinematic closure conditions and their analytical derivatives. This allows the solver to leverage the non-linear transmission effects inherent to closed-chain mechanisms, reducing peak actuator efforts and expanding their effective operating range. Unlike previous methods, our framework does not require serial approximations, enabling more accurate and efficient motion strategies. We also are able to generate the motion of more complex robots for which an approximate serial chain does not exist. We validate our approach through simulations and experiments, demonstrating superior performance in complex tasks such as rapid locomotion and stair negotiation. This method enhances the capabilities of current closed-loop robots and broadens the design space for future kinematic architectures
Poster: A microarchitectural signals analysis platform to craft Hardware Security Counters
No full textInternational audienceDetecting malicious software or hardware behavior during the operation of a computer system requires observables from one or more abstraction layers of the system. However, this abstraction tends to limit the ability to detect behavioral deviations, especially for attack classes that exploit vulnerabilities very close to the target hardware. Conversely, too low a level of abstraction tends to significantly increase the complexity of the system model, and therefore poses a number of difficulties for the extraction and selection of relevant observables for a given class of attack.In particular, processor performance counters have been used as an indirect means of observing microarchitecture behavior and detecting software attempting to exploit hardware vulnerabilities. In order to improve the various detection methods, we propose the construction of hardware metrics designed from the outset for security, by studying the correlation between signals from the microarchitecture and the various classes of attack in the literature, targeting both usual and industrial systems. By extension, this work aims to detect attacks originating from hardware Trojans, the latter having the effect of changing the behavior of a given microarchitecture
A wake-up strategy enabling GNSS-free NB-IoT links to sparse LEO satellite constellations
No full textInternational audienceThe latest release by the 3rd Generation Partnership Project (3GPP) defines how a non-terrestrial NB-IoT link may be set up between User Equipments (UE) on the ground and Low Earth Orbit (LEO) satellites equipped with Evolved Nodes B (eNB). However, a strong assumption is undertaken. Each UE must have Global Navigation Satellite Systems (GNSS) capabilities to properly pre-compensate the Doppler frequency shift and the propagation delay according to the time-varying relative motion of satellites. Additionally, although Release 18 accounts for discontinuous coverage by LEO satellites, the management of next passes over any spot on the Earth is undefined, thus affecting the system scalability. Remarkably, this contribution enables GNSS-free NB-IoT Direct-to-Satellite communications with sparse LEO satellite constellations. To do that, the UE periodically wakes up until it detects a satellite pass in its range. By listening to several NB-IoT beacons, the estimated Doppler curve is used to pre-compensate ongoing communications in frequency and time. Furthermore, the UE uses the standard information sent from the eNB, together with its own estimated location, to guess the next satellite pass without using GNSS. Simulation results reveal that the introduced wake-up strategy allows GNSS-free UEs to save more energy than if equipped with the most power-efficient GNSS chipsets surveyed in 3GPP specifications, promoting the broader deployment of IoT devices in remote and underserved areas.</div
Macromolecular interactions and geometrical confinement determine the 3D diffusion of ribosome-sized particles in live Escherichia coli cells
No full textInternational audienceThe crowded bacterial cytoplasm is composed of biomolecules that span several orders of magnitude in size and electrical charge. This complexity has been proposed as the source of the rich spatial organization and apparent anomalous diffusion of intracellular components, although this has not been tested directly. Here, we use biplane microscopy to track the 3D motion of self-assembled bacterial genetically encoded multimeric nanoparticles (bGEMs) with tunable size (20 to 50 nm) and charge (−3,240 to +2,700 e) in live Escherichia coli cells. To probe intermolecular details at spatial and temporal resolutions beyond experimental limits, we also developed a colloidal whole-cell model that explicitly represents the size and charge of cytoplasmic macromolecules and the porous structure of the bacterial nucleoid. Combining these techniques, we show that bGEMs spatially segregate by size, with small 20-nm particles enriched inside the nucleoid, and larger and/or positively charged particles excluded from this region. Localization is driven by entropic and electrostatic forces arising from cytoplasmic polydispersity, nucleoid structure, geometrical confinement, and interactions with other biomolecules including ribosomes and DNA. We observe that at the timescales of traditional single molecule tracking experiments, motion appears subdiffusive for all particle sizes and charges. However, using computer simulations with higher temporal resolution, we find that the apparent anomalous exponents are governed by the region of the cell in which bGEMs are located. Molecular motion does not display anomalous diffusion on short time scales and the apparent subdiffusion arises from geometrical confinement within the nucleoid and by the cell boundary