Real-time fluid simulations under smoothed particle hydrodynamics for coupled kinematic modelling in robotic applications

• Autores: Gabriel Camporredondo Diaz
• Directores de la Tesis: Ramón I. Barber Castaño (dir. tes.), Mathieu Legrand (codir. tes.), Lourdes Muñoz Gómez (codir. tes.)
• Lectura: En la Universidad Carlos III de Madrid ( España ) en 2021
• Idioma: español
• Tribunal Calificador de la Tesis: Carmen Martínez Arévalo (presid.), Luis Santiago Garrido Bullón (secret.), Benjamín Hernández Arreguín (voc.)
• Programa de doctorado: Programa de Doctorado en Ingeniería Eléctrica, Electrónica y Automática por la Universidad Carlos III de Madrid
• Enlaces
  • Tesis en acceso abierto en: e-Archivo
• Resumen

  • Although solids and fluids can be conceived as continuum media, applications of solid and fluid dynamics differ greatly from each other in their theoretical models and their physical behavior. That is why the computer simulators of each turn to be very disparate and case- oriented.

    The aim of this research work, captured in this thesis book, is to find a fluid dynamics model that can be implemented in almost real-time with GPU processing and that can be adapted to typically large scales found in robotic devices in action with fluid media. More specifically, the objective is to develop these fast fluid simulations, comprising different solid body dynamics, to find a viable time kinematic solution for robotics. The tested cases are: i) the case of a fluid in a closed channel flowing across a cylinder, ii) fluid flowing across a controlled flap, and iii), free surface fluid control while pouring vessels. The implementation of the former cases settles the formulations and constraints to the latter applications. The results will allow the reader not only to sustain the implemented models but also to break down the software implementation elements for better comprehension.

    In the main implementation a fast GPU-based fluid dynamics simulation is detailed. The results show that it can be used in real-time to allow robotics to perform a blind pouring task with a conventional controller and no special sensing systems nor knowledge-driven prediction models would be necessary.