Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
Thick Fuzzy Sets (TFSs) and Their Potential Use in Uncertain Fuzzy Computations and Modeling
No full textInternational audienceThis paper aims at proposing the concept of thick fuzzy sets (TFSs). A TFS is based on the joint use of thick sets (TSs) and -cuts concepts. A TFS is represented by a family of nested TSs. A TS is an uncertain set, which is represented by a pair of crisp sets (CSs). These CSs characterize the upper and lower bounds of the TS. Therefore, a TS can be regarded as an interval of CSs. In this framework, as a type-1 fuzzy set (T1FS) is regarded as a family of nested CSs, a TFS can be represented by a family of nested TSs. Furthermore, according to the vertical dimension, a TFS can be regarded as an interval with T1FS bounds. The potentialities of the TFS concept have been validated using application examples where a real-world application for modeling the zone explored by an underwater robot is given
CLIC: Curriculum Learning and Imitation for object Control in non-rewarding environments
No full textInternational audienceIn this paper we study a new reinforcement learning setting where the environment is non-rewarding, contains several possibly related objects of various controllability, where an apt agent Bob acts following its own goals, without necessarily providing helpful demonstrations, and where the objective of an agent is to learn to control objects individually. We present a generic discrete-state discrete-action model of such environments, and an unsupervised reinforcement learning agent called CLIC for Curriculum Learning and Imitation for Control to achieve the desired objective. CLIC selects objects to focus on when training and imitating by maximizing its learning progress. We show that CLIC can effectively observe Bob to gain control of objects faster, even if Bob is not explicitly teaching. Despite choosing what it imitates in a principled way, CLIC retains the natural ability to follow Bob when he provides ordered demonstrations. Finally, we show that compared with a non-curriculum based agent, when Bob controls objects that the agent cannot, or in presence of a hierarchy between objects in the environment, CLIC achieves faster mastery of the environment by ignoring non-reproducible and already mastered interactions with objects when imitating
Numerical modeling of liquid sloshing in flexible tank with FSI approach
No full textInternational audiencePurpose The paper aims to present a numerical modeling procedure for the analysis of liquid sloshing in a flexible tank subjected to an external excitation, with taking into account the effects of fluid–structure interaction (FSI). Design/methodology/approach A numerical model based on coupling a two-phase flow solver and an elastic solid solver is developed in OpenFOAM code. The Arbitrary Lagrangian–Eulerian formulation is adopted for the two-phase Navier–Stokes equations in a moving domain. The volume of fluid (VOF) method is applied for the air–liquid interface tracking. The finite volume method is used for the spatial discretization of both the fluid and the structure dynamics equations. The FSI coupling problem is solved by an explicit coupling scheme. The model is validated for linear and nonlinear sloshing cases. Then, it is used to analyze the effects of the liquid sloshing on the dynamic response of the tank and the effects of the tank flexibility on the liquid sloshing. Findings The obtained results show that the flexibility of the tank walls amplifies the amplitude of the sloshing and increases the fluctuation period of the air–liquid interface. Furthermore, it is found that the bending moment acting on the tank walls may be underestimated when rigid walls assumption is adopted as usually done in sloshing tank modeling. Also, tank walls flexibility causes a phase shift in the free surface dynamic response. Originality/value A review of previous studies on liquid sloshing in flexible tanks revealed that FSI effects have not been clearly and comprehensively analyzed for large-amplitude liquid sloshing. Many physical and numerical aspects of this problem still require clarifications and enhancements. The added value of the present work and its originality lie in the investigation of large-amplitude liquid sloshing in flexible tanks by using a staggered coupling approach. This approach is carried out by an original combination of a linear solid solver with a two phase fluid solver in OpenFOAM code. In addition, FSI effects on some response quantities, identified and analyzed herein, have not been found in the previous works
TeachMyAgent: a Benchmark for Automatic Curriculum Learning in Deep RL
No full textInternational audienceTraining autonomous agents able to generalize to multiple tasks is a key target of Deep Reinforcement Learning (DRL) research. In parallel to improving DRL algorithms themselves, Automatic Curriculum Learning (ACL) study how teacher algorithms can train DRL agents more efficiently by adapting task selection to their evolving abilities. While multiple standard benchmarks exist to compare DRL agents, there is currently no such thing for ACL algorithms. Thus, comparing existing approaches is difficult, as too many experimental parameters differ from paper to paper. In this work, we identify several key challenges faced by ACL algorithms. Based on these, we present TeachMyAgent (TA), a benchmark of current ACL algorithms leveraging procedural task generation. It includes 1) challengespecific unit-tests using variants of a procedural Box2D bipedal walker environment, and 2) a new procedural Parkour environment combining most ACL challenges, making it ideal for global performance assessment. We then use TeachMyAgent to conduct a comparative study of representative existing approaches, showcasing the competitiveness of some ACL algorithms that do not use expert knowledge. We also show that the Parkour environment remains an open problem. We open-source our environments, all studied ACL algorithms (collected from open-source code or re-implemented), and DRL students in a Python package available at https://github. com/flowersteam/TeachMyAgent
A numerical strategy to identify the FSW process optimal parameters of a butt-welded joint of quasi-pure copper plates: modeling and experimental validation
No full textInternational audienceDetermining the optimal parameters of the friction stir welding (FSW) process, which are suitable for a given joint configuration, remains a great challenge and is often achieved through extremely time-consuming and costly experimental investigations. The present paper aims to propose a strategy for the identification of the optimal parameters for a butt-welded joint of 3-mm thick quasi-pure copper plates. This strategy is based on FEM (finite element method) simulations and the optimal temperature that is supposedly known. A robust and efficient finite element model that is based on the Coupled Eulerian-Lagrangian (CEL) approach has been adopted and a temperature-dependent friction coefficient has been used. Besides, the mass scaling technique has been used to significantly reduce the simulation time. The thermo-mechanical behavior of the butt-welded joint was modeled using a Johnson-Cook plasticity model that was identified through lab tests at different temperatures. The results of the parametric study help to define the numerical surface response, and based on this latter one can find the optimal parameters, advancing (va) and rotational (vr) speeds, of the FSW process. This numerical surface response has been validated with good agreement between the numerical prediction of the model and the experimental results. Furthermore, experimental investigations involving x-ray radiography, digital image correlation method, and fracture surface analysis have helped a better understanding of the effects of FSW parameters on the welded joint quality
Properties of the pressure field in highly nonlinear free surface flows with critical jet
No full textInternational audienceIn this note we examine fluid violent kinematics in a plunging breaker. The fluid motion is computed in the frame of the potential flow theory. The fluid kinematics is basically generated by forcing the motion of a rectangular tank, starting from rest. When a sufficient level of energy is injected in the fluid, the free surface has highly nonlinear behavior, here a plunging breaker. In the vicinity of the main tip crest, a sharp corner (cusp) appears along the surface of the barrel. The appearance of this critical jet is described and discussed in terms of the spatial and temporal variations of the pressure field. In the present case a local pressure maximum is captured that follows a continuous decreasing pressure gradient in a region of positive Gaussian curvature of the pressure. It is also shown that, at the free surface before the appearance of the critical jet, there is a strong correlation between the change of sign of the Gaussian curvature of the pressure on the one hand and the radius of curvature of the free surface profile on the other hand
Time-resolved study of laser emission in nitrogen gas pumped by two near IR femtosecond laser pulses
No full textInternational audienceThe time profile of the lasing signal at 391.4 nm emitted by a weakly ionized gas of nitrogen molecules at low pressure is measured under double excitation with intense femtosecond laser pulses at 800 nm. An abrupt decrease of the emission occurs at the time of arrival of the second pulse. It is explained by a transfer of population from ground to first excited ionic level and by a disruption of coherence, terminating the conditions for lasing in a V-scheme without population inversion
On the proof of recursive Vogler algorithm for multiple knife-edge diffraction
No full textInternational audienceWe consider the problem of multiple knife-edge diffraction estimation which is a fundamental task in many wireless communication applications. So far, one of the most accurate methods for this problem is the Vogler one whose recursive implementation is efficient to reduce the high computational complexity of the direct one. However, in the original report, Vogler only presented the final result of the recursive algorithm without a rigorous mathematical proof, thus making the method difficult to understand and implement in practice. To tackle this shortcoming, we first analyze the mathematical structure of the problem and then present a formal proof of the result. To gain intuition of the proof and the key steps, we provide a simplified study case of four knife-edges. The insight from our proposed analysis and proof can be used to obtain a comprehensive interpretation, initiate a practical implementation and develop new efficient algorithms with similar structure