PhD Repository: Development of an Intelligent Experimental Setup for the Measurement and Manipulation of the Hydrodynamic State of a Basin with Submerged Bodies.
The initial phase of this research focuses on the dimensioning of sensors intended for use in aquatic environments to estimate the pressure generated by a swimming fish. This process involves utilizing a Julia library developed by Gabriel D. Weymouth, named WaterLily. WaterLily enables the generation of a fish's desired shape and the parametrization of its speed and motion in a fluid. Consequently, access to various simulation characteristics, such as pressure and velocity maps, is facilitated.
The conversion table then allows the transformation of dimensionless data into real-world values. This table is derived from Gabriel D. Weymouth's work, specifically from his Wiki on the Lilypad library.
An Automated and Highly Dynamic Digital Twin Design with YOLOv8 for Hydrodynamic Studies on Living Animals.
The subsequent phase of this research entails generating Computational Fluid Dynamics (CFD) simulations closely aligned with real experiments. This involves extracting the fish from experiment videos, utilizing key features to construct a digital twin. The chosen CFD software for this purpose is LilyPad, also developed by Gabriel Weymouth. LilyPad, unlike other alternatives, employs point-based methods rather than functions for shape generation. A pipeline is developed to extract fish features using YOLOv8, automating the generation of a corresponding simulation. The fidelity of the digital twin is then assessed by comparing real pressure measurements from the sensors with those extracted from the simulation.