From Fluid–Structure Modeling to Reinforcement Learning: Controlling Micro-Swimmers in Complex Environments

Image

Laetitia Giraldi

Mercredi 12 novembre 2025 à 15h00, salle des séminaires IRPHE

Abstract: Micro-swimming, which describes the propulsion of microorganisms or micro-robots immersed in complex fluids, raises fundamental questions at the interface between fluid mechanics, applied mathematics, and robotics. 
In this talk, I will address issues related to the mathematical modeling, control, and optimization of such active systems. One of the main challenges lies in accurately computing their dynamics, which is particularly delicate due to the underlying hydrodynamic interactions. I will present the mathematical model of a digital twin of a magnetic and elastic micro-robot, whose dynamics are solved using FEM–ALE (Finite Element Method – Arbitrary Lagrangian–Eulerian) techniques, allowing for a detailed description of fluid–structure interactions. In the second part of the talk, I will show how a global control, learned through reinforcement learning, can be used to efficiently guide a swarm of swimmers in a confined environment, enhance collective transport.

Laetitia Giraldi, Inria Sophia Antipolis