Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
Rupture d’interface rigide-élastomère sous chargement complexe
No full textElastomers reinforced by textile fibers or metallic wires are common in numerous industrial applications (tires, driving belts, conveying belts, and pneumatic springs). The adhesion between the reinforcement and the elastomer matrix must be ensured to guarantee the structural integrity of the product under various mechanical loads. A reliable characterization of the crack propagation resistance of the interface is thus necessary, motivating the development of new mechanical tests. Firstly, an experimental setup of crack propagation between a metal wire and a natural black carbon rubber is used to charactérize the dissipated energy during the interfacial fracture. The test, developed from a precedent thesis, is first improved and the experimental methodology is updated. Analytical and numerical models allow the validation of the test kinematics, experimental conditions of lubrication, and the computation of the adhesion energy. Secondly, the test is derived using three methods, permitting the characterization of the adhesion between the reinforcement and the matrix under other loadings. Hence, the crack mode mixity, the crack propagation velocity, or the matrix pre-softening alters the global dissipation during the interface fracture. Those new methods offer an adhesion characterization in closer conditions to the industrial applications.Les élastomères armés de fibres textiles ou de fils métalliques se retrouvent dans de nombreuses applications industrielles (pneus, courroies d'entraînement, bandes de convoyage, tuyaux, ressorts pneumatiques). L'adhésion du renfort à la matrice élastomère doit être assurée pour garantir l'intégrité structurelle du produit sous divers chargements mécaniques. Une caractérisation fiable de la résistance à la fissuration de l'interface est donc nécessaire, ce qui motive le développement de nouveaux essais mécaniques. Dans un premier temps, un dispositif expérimental de fissuration entre un fil métallique et un caoutchouc naturel chargé de noir de carbone est utilisé pour caractériser la dissipation d'énergie lors de la rupture interfaciale. L’essai, issu d’une précédente thèse, est tout d’abord amélioré et la méthodologie expérimentale consolidée. Des modélisations analytiques et numériques permettent de valider la cinématique de l’essai, les conditions expérimentales de lubrification et le calcul de l’énergie d’adhésion. Dans un second temps, l’essai est décliné selon trois méthodes permettant de caractériser l’adhésion du renfort à la matrice sous d’autres chargements. Ainsi, il apparaît que la mixité du chargement en pointe de fissure, la vitesse de propagation ou la pré-accommodation du caoutchouc influent sur l’énergie globale dissipée pendant la rupture de l’interface. Ces nouvelles méthodes offrent alors une caractérisation de l’adhésion dans des conditions plus proche des applications industrielles
Stick to your role! Stability of personal values expressed in large language models
No full textInternational audienceThe standard way to study Large Language Models (LLMs) through benchmarks or psychology questionnaires is to provide many different queries from similar minimal contexts (e.g. multiple choice questions). However, due to LLM’s highly context-dependent nature, conclusions from such minimal-context evaluations may be little informative about the model’s behavior in deployment (where it will be exposed to many new contexts). We argue that context-dependence should be studied as another dimension of LLM comparison alongside others such as cognitive abilities, knowledge, or model size. In this paper, we present a case-study about the stability of value expression over different contexts (simulated conversations on different topics), and as measured using a standard psychology questionnaire (PVQ) and behavioral downstream tasks. We consider 21 LLMs from six families. Reusing methods from psychology, we study Rank-order stability on the population (interpersonal) level, and Ipsative stability on the individual (intrapersonal) level. We explore two settings: with and without instructing LLMs to simulate particular personalities. We observe similar trends in the stability of models and model families—Mixtral, Mistral, GPT-3.5 and Qwen families being more stable than LLaMa-2 and Phi—over those two settings, two different simulated populations, and even on three downstream behavioral tasks. When instructed to simulate particular personas, LLMs exhibit low Rank-Order stability, and this stability further diminishes with conversation length. This highlights the need for future research directions on LLMs that can coherently simulate a diversity of personas, as well as how context-dependence can be studied in more thorough and efficient ways. This paper provides a foundational step in that direction, and, to our knowledge, it is the first study of value stability in LLMs. The project website with code is available at https://sites.google.com/view/llmvaluestability
AI-driven Automated Discovery Tools Reveal Diverse Behavioral Competencies of Biological Networks
No full textInternational audienceMany applications in biomedicine and synthetic bioengineering depend on the ability to understand, map, predict, and control the complex, context-sensitive behavior of chemical and genetic networks. The emerging field of diverse intelligence has offered frameworks with which to investigate and exploit surprising problem-solving capacities of unconventional agents. However, for systems that are not conventional animals used in behavior science, there are few quantitative tools that facilitate exploration of their competencies, especially when their complexity makes it infeasible to use unguided exploration. Here, we formalize and investigate a view of gene regulatory networks as agents navigating a problem space. We develop automated tools to efficiently map the repertoire of robust goal states that GRNs can reach despite perturbations. These tools rely on two main contributions that we make in this paper: (1) Using curiosity-driven exploration algorithms, originating from the AI community to explore the range of behavioral abilities of a given system, that we adapt and leverage to automatically discover the range of reachable goal states of GRNs and (2) Proposing a battery of empirical tests inspired by implementation-agnostic behaviorist approaches to assess their navigation competencies. Our data reveal that models inferred from real biological data can reach a surprisingly wide spectrum of steady states, while showcasing various competencies that living agents often exhibit, in physiological network dynamics and that do not require structural changes of network properties or connectivity. Furthermore, we investigate the applicability of the discovered "behavioral catalogs" for comparing the evolved competencies across classes of evolved biological networks, as well as for the design of drug interventions in biomedical contexts or for the design of synthetic gene networks in bioengineering. Altogether, these automated tools and the resulting emphasis on behavior-shaping and exploitation of innate competencies open the path to better interrogation platforms for exploring the complex behavior of biological networks in an efficient and cost-effective manner
Cooperative control of environmental extremes by artificial intelligent agents
No full textInternational audienceHumans have been able to tackle biosphere complexities by acting as ecosystem engineers, profoundly changing the flows of matter, energy and information. This includes major innovations that allowed to reduce and control the impact of extreme events. Modelling the evolution of such adaptive dynamics can be challenging given the potentially large number of individual and environmental variables involved. This paper shows how to address this problem by using fire as the source of extreme events. We implement a simulated environment where fire propagates on a spatial landscape, and a group of artificial agents learn how to harvest and exploit trees while avoiding the damaging effects of fire spreading. The agents need to solve a conflict to reach a group-level optimal state: while tree harvesting reduces the propagation of fires, it also reduces the availability of resources provided by trees. It is shown that the system displays two major evolutionary innovations that end up in an ecological engineering strategy that favours high biomass along with the suppression of large fires. The implications for potential A.I. management of complex ecosystems are discussed
Silicone-Based Haptic Interfaces: Enhancing Multimodal Interactions through Pneumatic Tactile Feedback
No full textInternational audienceThis paper explores the potential of pneumatic haptic interfaces in enhancing human-computer interaction. We present two projects: one augmenting movie experiences with emotion-synchronized haptic feedback, and another integrating pneumatic interfaces into steering wheels for improved driver takeover in autonomous vehicles. The movie experience project demonstrated enhanced emotional engagement, while the automotive application showed improved safety and user trust. These applications highlight the versatility of pneumatic haptic technology across entertainment and safety contexts. We discuss the advantages of the Baromorph technique and outline future research directions, including comparative studies with vibrotactile feedback and machine learning approaches. This work contributes to the development of more insightful and emotionally intelligent interactive systems
Les bénéfices d'un entrainement cognitif informatisé et individualisé chez les personnes âgées
No full textInternational audienceThis randomized controlled trial included 50 healthy older adults, divided into AI-basedindividualized adjustment ( ZPDES ) and traditional staircase (control) groups. We assessedtask performance progression, cognitive transfer across seven tasks, interindividual differ-ences, and subjective experiences via questionnaires. Pre-post comparisons revealed greater training benefits in the MOT task for the ZPDES group. Specific task improvements were noted only in the control group, but no differences were observed at a latent level. Both groups exhibited non-linear intra-training progress: control participants improved initially and then plateaued, while ZPDES participants showed consistent progress throughout the two weeks of training. For both conditions, interindividual differences in prior MOT performance significantly influenced baseline performance in the first training session, but no significant differences were found in performance changes. Differences in training trajectories led to varied subjective experiences: cognitive load decreased more over time for the control group, indicating ZPDES was more demanding. Participants in the ZPDES group reported lower intrinsic and extrinsic motivation but a higher sense of competence. In sum, the ZPDES condition demonstrated greater post-training MOT performance, consistent intra-training progress, and higher competence, despite being more demanding. Consequently, this study discusses the implications of our approach on training benefits, experience, and engagement, and proposes improvements to the ZPDES algorithm to better address interindividual differences in cognitive aging
ROUGH PATHS AND SYMMETRIC-STRATONOVICH INTEGRALS DRIVEN BY SINGULAR COVARIANCE GAUSSIAN PROCESSES
No full textInternational audienceWe examine the relation between a stochastic version of the rough path integral with the symmetric-Stratonovich integral in the sense of regularization. Under mild regularity conditions in the sense of Malliavin calculus, we establish equality between stochastic rough path and symmetric-Stratonovich integrals driven by a class of Gaussian processes. As a by-product, we show that solutions of multi-dimensional rough differential equations driven by a large class of Gaussian rough paths they are actually solutions to Stratonovich stochastic differential equations. We obtain almost sure convergence rates of the first-order Stratonovich scheme to rough paths integrals in the sense of Gubinelli. In case the time-increment of the Malliavin derivative of the integrands is regular enough, the rates are essentially sharp. The framework applies to a large class of Gaussian processes whose the second-order derivative of the covariance function is a sigma-finite non-positive measure on + off diagonal
Identification and listing of operation research problems in the framework of heterogeneous robotic swarms in System-of-Systems: a path for consistent research targets
No full textInternational audienceRecent events in Ukraine revealed the progression of robotic systems on battlefield, both aerial and ground systems that answer various capabilities. Heterogeneous unmannedsystems acting in swarms are being tested on field and areevaluated through defense challenges. The capabilities and system views are exposed using a set of NATO Architecture Framework (NAF) 3.1 views. Based on field experience, we expose a selection of interconnected views of our systems, capabilities and operational activities in unstructured environment. The core idea of this paper is to introduce operation research (OR) problems that could be tackled to our views generation, so that any heterogeneous robotic swarm (HRS) architecture can be seen as a system-of-systems. Finally, by introducing those system-of-system architecting activities, we share insight on how HRS design is a highly multi-physical compromise to reach, that can only be improved over time if all multi-physical aspects of the system are considered
Model Free Deep Deterministic Policy Gradient Controller for Setpoint Tracking of Non-minimum Phase Systems
No full textInternational audienceDeep Reinforcement Learning (DRL) techniques have received significant attention in control and decision making algorithms. Most applications involve complex decision making systems, justified by the algorithms' computational power and cost. While model-based versions are emerging, model-free DRL approaches are intriguing for their independence from models, yet they remain relatively less explored in terms of performance, particularly in applied control. This study conducts a thorough performance analysis comparing the data-driven DRL paradigm with a classical statefeedback controller, both designed based on the same cost (reward) function of the linear quadratic regulator (LQR) problem. Twelve additional performance criteria are introduced to assess the controllers' performance, independent of the LQR problem for which they are designed. Two Deep Deterministic Policy Gradient (DDPG)-based controllers are developed, leveraging DDPG's widespread reputation. These controllers are aimed at addressing a challenging setpoint tracking problem in a Non-Minimum Phase (NMP) system. The performance and robustness of the controllers are assessed in the presence of operational challenges, including disturbance, noise, initial conditions, and model uncertainties. The findings suggest thatthe DDPG controller demonstrates promising behavior under rigorous test conditions. Nevertheless, further improvements are necessary for the DDPG controller to outperform classical methods in all criteria. While DRL algorithms may excel in complex environments owing to the flexibility in the rewardfunction definition, this paper offers practical insights and a comparison framework specifically designed to evaluate these algorithms within the context of control engineering
Fluid-Structure Interactions Response of a Composite Hydrofoil Modelled With 1D Beam Finite Elements
No full textInternational audience_ In this paper, the hydroelastic response of a NACA0015 composite hydrofoil is studied experimentally and numerically. The foil is made of composite materials with fibers not aligned with the span of the foil, which results in the occurence of a bend-twist coupling in the material. Computations are performed using a partitioned approach. The flow problem is solved using a boundary element method. The structural response of the foil is modelled with two different finite element models. In the first one, the foil is modelled with 2D shell and 3D solid finite elements and in the second model, the foil is modelled with 1D beam finite elements. The experiments are conducted in an open circulation water channel. Hydrodynamic forces and structural displacements are measured for several angles of attack, free stream velocities and submergence depth. This paper shows that the mechanical behaviour of a composite hydrofoil submitted to hydrodynamic loads can be modelled with 1D beam finite elements. This model gives results very similar to a finite element analysis realized with 2D shell and 3D solid finite elements, which are commonly used to model composite structures. The present work also shows that the experimental results can be well predicted by numerical simulations, but it requires a precise modeling of the bend-twist coupling in the materials constituting the foil. Keywords Hydrofoil; Equivalent Beam; Fluid-Structure Interactions; Composite; Bend-Twist Couplin