Archive ouverte HAL-LAAS
Not a member yet
12189 research outputs found
Sort by
Towards a Fully Integrated Refractometric Sensor in Silicon Nitride: Integrated Spectrometer Design
No full textInternational audienc
Stereolithography 3D printing method for multi-material hydrogel 2D photo-patterning in a microfluidic chip
No full textInternational audienceWe present a novel and straightforward method using a standard stereolithography (SLA) 3D printer for highresolution (20 μm x-y resolution), multi-material 2D hydrogel photo-patterning directly within a microfluidic chip. The process involves sequential injections of photosensitive hydrogel into a transparent microfluidic chip coupled with sequential direct laser writing by the printer through point-by-point photopolymerization. Our approach integrates a custom miniaturized syringe pump system into the SLA printer, thereby enabling fluid management and sequential injection of different photosensitive hydrogels directly into the microfluidic environment between each laser writing sequence. This technique enables the fabrication of intricate, multi-material hydrogel patterns (e.g., PEGDA and HAMA) with high spatial resolution over areas spanning several square millimeters. Future developments will focus on expanding the range of biomaterials and incorporating cell-laden hydrogels to facilitate the creation of biologically relevant microenvironments on chip.This study opens new possibilities for high-resolution, multi-material hydrogel patterning in microfluidics and offers a valuable platform for advancing research in microsystems engineering.</p
Inverse optimal control reveals square of muscle power minimisation in post‐stroke gait
No full textInternational audienceThis study investigated how the nervous system controls muscles during post-stroke gait. Muscle forces were obtainedin a previous study using an EMG-driven musculoskeletal model. The closest predictions were obtained in the inverseoptimal control by minimising the sum of muscle powers squared. This objective function, though rarely used in staticoptimisation, may be especially important for understanding abnormal gait patterns caused by conditions like spasticity
Integrating Soft and Rigid Elements in the Design of Tools for Studying Human Tactile Communication for Robot Interaction Control
No full textInternational audienceTactile communication plays a major role in interhuman relations and could play an important role in human-robot interaction. However, many challenges still exist that prevent robots from mastering the skills necessary for intelligent and proactive communication via physical touch. In order to contribute to enabling robots to engage in proactive tactile communication with humans, we explore how to jointly address some of the related challenges at the control and hardware levels. We present the iterative prototype development of a hand-sensor attachment for collecting database of inter-human tactile communication samples in order to extract patterns that can be fed into whole-body robot controller as a force tracking tasks to replicate the communication practices. In order to enable the robot to interact in a more human-like fashion, we also develop the human finger inspired end-effector design that incorporates soft material and rigid sensor attachment modules
Differentiable ray-tracer used for coded hyperspectral systems co-design
No full textInternational audienceWe developed a differentiable ray-tracing optics simulator based on PyTorch and a framework previously designed for systems with axial symmetry.We extended this framework to non-symmetric systems and a wider variety of optical elements and systems.We believe that such a tool is particularly well suited to a co-design approach.We use this simulator to model a Coded Aperture Snapshot Spectral Imager (CASSI) for coded hyperspectral imagery.These systems employ a coded aperture (mask) to modulate the 3D hyperspectral scene, resulting in a 2D acquisition.We then adapted existing unrolling reconstruction algorithms to solve the inverse problem of recovering scenes from coded acquisitions.We demonstrated that CASSI systems using the same mask but inducing different distortions can reconstruct scenes with similar performance, provided that the distortions are correctly modeled and taken into account.We can therefore state that the coded information is constant whatever the distortions on such an imager, but that it can depend on the mask.Thanks to the differentiability of our simulator, all its parameters—especially the mask—can be optimized with respect to a downstream loss function, leading to co-design. As reconstruction times are prohibitive in our high-dimensional use case, we attempt to bypass reconstruction by optimizing the mask to maximize the information gain that quantifies the information of the scene encoded in the acquisition.We hypothesize that an increase in information gain will lead to better processing performance, with a significant reduction in optimization time
Synchronization of Multi-Agent Hybrid Systems With Synchronous State-Dependent Jumps
No full textInternational audienceIn this letter, we investigate the problem of synchronization of multi-agent systems described by hybrid non-linear input affine dynamics. We consider networks described by undirected connected graphs without leader. We present a set of sufficient conditions based on an LMI approach in order to design a state-feedback distributed control law. Then, by exploiting the properties of the graph incidence matrix we provide an optimization of the tuning parameters. The incidence matrix properties also allow us to construct a Lyapunov function to establish synchronization of the hybrid network
Réseaux résonnants en cavité sur niobate de lithium pour spectroscopie haute résolution
No full textDéveloppement de filtres à réseaux résonnants en cavité intégrés sur niobate de lithium pour des applications d'imagerie hyperspectrale haute résolutio
Log Probability Tracking of LLM APIs
No full textWhen using an LLM through an API provider, users expect the served model to remain consistent over time, a property crucial for the reliability of downstream applications and the reproducibility of research. Existing audit methods are too costly to apply at regular time intervals to the wide range of available LLM APIs. This means that model updates are left largely unmonitored in practice. In this work, we show that while LLM log probabilities (logprobs) are usually non-deterministic, they can still be used as the basis for cost-effective continuous monitoring of LLM APIs. We apply a simple statistical test based on the average value of each token logprob, requesting only a single token of output. This is enough to detect changes as small as one step of fine-tuning, making this approach more sensitive than existing methods while being 1,000x cheaper. We introduce the TinyChange benchmark as a way to measure the sensitivity of audit methods in the context of small, realistic model changes
On Regional Sampled-Data Control of Nonlinear Time-Delay Systems with Input Saturation Constraints
No full textInternational audienceIn this paper, we deal with the stabilization problem of nonlinear time-delay systems affected by input saturation constraints via sampled-data dynamic output feedback controllers. Nonlinear time-delay systems not necessarily affine in the control input are studied. A new approach for the design of sampled-data dynamic output feedback controllers is provided taking into account the effects of input saturation constraints. In particular, it is shown that the digital implementation via suitable approximation functions of dynamic output steepest descent feedback (continuous or not) with amplitude bounds yields the practical stability, with arbitrarily small final target ball, of the related sampled-data closed-loop system limited to suitably small regions. The case of time-varying sampling intervals and the stability analysis of the inter-sampling system behavior are included in the theory here developed. The provided results can be applied also for nonlinear delay-free systems which are here addressed as a special case. An example is presented which validates the theoretical results
Optimizing ZIF-8 membrane growth on top of semiconductive Ga-doped ZnO sensitive layers
No full textInternational audienceFunctionalizing ZnO-based chemiresistive sensor surfaces with a ZIF-8 layer, which acts as a permselective membrane, is a well-established strategy to enhance sensor selectivity. This study examines the key factors influencing the conversion process, including the physicochemical properties of solvents or solvent mixtures and the thermal pretreatment, of Ga-doped ZnO (ZnO : Ga). We have evidenced that the polarity, viscosity, and interfacial tension of the solvent significantly affect the dissolution of ZnO : Ga and the crystallization of ZIF-8. Methanol–water mixtures were found to effectively control the conversion process, with a 3 : 1 MeOH/H2O ratio being optimal for producing continuous ZIF-8 membranes. Additionally, it has been demonstrated that annealing can greatly enhance the reactivity of the oxide. However, while it enhances the dissolution of ZnO : Ga, excessively high temperatures can lead to over-dissolution, which hinders ZIF-8 formation. These insights are crucial for optimizing ZIF-8 layers on ZnO : Ga, paving the way for the development of highly selective chemiresistive sensors