Entité de rattachement
Le CEA est un acteur majeur de la recherche, au service des citoyens, de l'économie et de l'Etat.
Il apporte des solutions concrètes à leurs besoins dans quatre domaines principaux : transition énergétique, transition numérique, technologies pour la médecine du futur, défense et sécurité sur un socle de recherche fondamentale. Le CEA s'engage depuis plus de 75 ans au service de la souveraineté scientifique, technologique et industrielle de la France et de l'Europe pour un présent et un avenir mieux maîtrisés et plus sûrs.
Implanté au cœur des territoires équipés de très grandes infrastructures de recherche, le CEA dispose d'un large éventail de partenaires académiques et industriels en France, en Europe et à l'international.
Les 20 000 collaboratrices et collaborateurs du CEA partagent trois valeurs fondamentales :
• La conscience des responsabilités
• La coopération
• La curiosité
Référence
2024-30865
The postdoc consists in simulating the contact interaction of deformable bodies modeled as finite elements with surrogate models in the form of Neural Networks.
The postdoc will contribute to bring currently offline simulations into the world of interactive simulations by coupling well-established numerical methods (typically, Finite Elements) which are ”slow”, to emerging surrogate models (seen as Reduced-Order Models) involving Neural Networks.
The first objective is to provide a simulation that couples a finite-element beam and another beam whose dynamic behaviour is encoded in a NN. The interest of this system is that it represents a wide class of similar problems, while being simple to set up: due to the one-dimensional nature of a beam, the sequential numbering of the nodes is trivial.The Interactive Simulation Laboratory (CEA List, France) already disposes of :
1) a NN trained to simulate the dynamics of a beam and
2) in-house beam finite element software XDE Physics.
The second objective is to produce a usable demonstrator, in the form of a Unity (or Unreal) program relying on our VirtuoseTM haptic device, that emulates the interaction of a robot (modeled with few degrees of freedom) with a deformable three-dimensional body. Such models find applications for tele-surgery (deformable body: an organ), manufacturing (deformable body: e.g. food), nuclear waste (deformable body: bituminous material), etc.
More than the application it targets, the real purpose of the demonstrator is to prove that the challenges of the coupling are raised, which is the end objective.
The coupling equations Coupling Finite Elements and Neural Networks will require to get familiar with numerical methods dedicated to non-smooth systems (ie. systems interacting through contact in this context). In the non-smooth settings of Measure Differential Inclusions, the coupling between two dynamical systems coupled through contact can be written in a very general form.
Keywords : Neural networks, surrogate models, co-simulation, dynamics, finite elements, interactive simulation
PhD in Applied Mathematics, Neural Networks or Computational methods for mechanics
Je me crée un espace candidat
