Offre en lien avec l’Action/le Réseau : – — –/– — –

Laboratoire/Entreprise : LaMcube, en collaboration avec CRIStAL
Durée : 6 mois
Contact : jean-baptiste.colliat@univ-lille.fr
Date limite de publication : 2024-01-31

Contexte :
Ce stage se place dans le contexte d’une nouvelle collaboration entre le LaMcube, laboratoire de mécanique de l’université de Lille., et CRIStAL, laboratoire d’informatique, autmatique et traiement de signal de l’Université de Lille.

L’objectif est d’explorer l’utilisation de l’apprentissage par renforcement pour le test de matériaux.

Sujet :
The goal of this internship is to design, implement and test a reinforcement learning agent able to control a material testing machine. This machine is used to perform research on materials at the LaMcube lab. The reinforcement learning part of the internship will be handle in collaboration with team Scool at CRIStAL/Inria.

Mechanics of materials aims to understand, model and optimize the mechanical response of industrially relevant materials. Here, the scale of observation as well as the size of the specimens are the keystones in order to build an accurate identification strategy. Major improvements have been made during the last four decades, mainly thanks to the renewal of measurement techniques. Still, several material properties and field values are difficult to measure directly. This is especially true for the interfaces. Moreover, the search for adaptive loading paths able to activate specific fine scale mechanisms is of the greatest interest, regardless of the material.
During this internship, we aim to develop a novel experimental-numerical technique in order to determine such quantities of interest by selecting the optimal macroscopic multiaxial loading paths. Reinforcement learning is coupled with material testing to attain this goal. The objective is to explore several RL algorithms in order to train an agent to control the material testing machine. A simulation environment based on the Finite Element Method will be used to train the RL agent.

(See the attached pdf for pictures.)

Profil du candidat :
Strong knowledge in reinforcement learning.
Knowledge in mechanics is a plus.
Ability to communicate and work on an interdisciplinary project.
Autonomous, able to propose original and realistic ideas.
Interested in practical applications of RL.

Formation et compétences requises :
Master d’informatique avec une spécialisation en apprentissage automatique.

Adresse d’emploi :
Cité scientifique, Villeneuve d’Ascq.

Document attaché : 202311281450_RL_intern_CRISTAL_LAMCUBE.pdf