Forward and inverse problems in fluid-body interactions: immersed simulation methods and data-driven reduced-order models

Fluid-structure interactions involving flexible and actuated structures present significant opportunities to advance biologically inspired swimming and morphing airfoils. Two major challenges in this field are achieving efficient, high-fidelity simulations of these phenomena in three dimensions and solving inverse problems for optimal design and control. Addressing the latter is particularly demanding in robotic systems, where the vast parameter space enabled by modern smart materials and structures makes brute-force optimization methods impractical. In this talk, I will discuss recent work tackling forward and inverse problems in fluid-structure interaction. First, I will outline advancements in 3D simulations of fluid flows interacting with complex, moving boundaries, utilizing high-order immersed methods and adaptive grids to improve efficiency and accuracy. Second, I will discuss a data-driven strategy for solving inverse-design problems, using a machine learning model to predict loading distributions on deforming bodies. These contributions offer complementary approaches to scientific analysis and practical application of fluid-structure interactions.