Simulated reality: physics-based mixed and augmented reality for intelligence augmentation
- Autores: Alberto Badías Herbera
- Directores de la Tesis: David González Ibáñez (dir. tes.), Icíar Alfaro Ruiz (codir. tes.)
- Lectura: En la Universidad de Zaragoza ( España ) en 2020
- Idioma: español
- Tribunal Calificador de la Tesis: Francisco Javier Oliver Olivella (presid.), Estefanía Peña Baquedano (secret.), Pierre Ladeveze (voc.)
- Programa de doctorado: Programa de Doctorado en Ingeniería Mecánica por la Universidad de Zaragoza
- Materias:
- Enlaces
- Tesis en acceso abierto en: Zaguán
- Resumen
Simulated reality is a new paradigm that allows computers to understand and reproduce the physical phenomena that occur in their environment. By physical phenomena we mean the static and dynamic description of the changes that happen around us: rigid body movements, solid objects that deform, liquids that flow or change their state, gases that disperse, electromagnetic waves that propagate or even the joint of all the previous effects. Any phenomenon that can be simulated by means of a computer has a place within simulated reality, which consists of reproducing these interactions in a virtual way (inside a computer) to be subsequently shown in real time to a user. Our work consists in creating a human-centered tool that allows the user to observe the behavior of all these physical phenomena from a scientific point of view, that is, perceiving information about stresses, strains, velocities, flow rates or intensities with their specific values at the same time they are happening. It is about providing information that the user cannot perceive directly with his senses (intelligence augmentation), putting to his service a series of tools that allow him to make decisions with a greater capacity of knowledge, or simply to observe the dynamic phenomena promoting higher comprehension.
In general, the proposed tool requires a great capacity of computation to describe all these phenomena. The simulation of physical problems usually requires large computational resources that are far from solving the equations in real time. Complex models with non-linearities and coupled phenomena usually appear in this type of problems. This is why we use model order reduction techniques, to reduce the complexity of the models and evaluate them at the same speed as real phenomena using data assimilation procedures, to later show the results to the user. The most common methods of dimensionality reduction require to preprocess the solutions, carried out off-line, but allowing an on-line evaluation that meets time requirements. This type of off-line processing usually include multiparametric solutions that cover a wide range of solutions in order to create physically consistent estimations in the on-line evaluation. In addition, dimensionality reduction methods project the data to a new space that is, usually, more efficient, which translates into the compression of the data to reduce the storage space, at the same time very fast evaluations for specific parametric values are assured.
In order to favour the assimilation of all this amount of information by the user, we use mixed reality tools and visual devices. We show the information by creating augmented video sequences with physical information coming from the simulations, allowing an interactive and simple visualization of the results. Since the communication is done by visual path, we are forced to work at a frequency around 30-60 Hz, since these are the standard refresh frequencies on the common devices.
In this thesis a complete framework has been developed covering all the necessary pieces to create a simulated reality system, including new methodologies in model order reduction methods using non-linear projections, the application of existing order reduction methods on new physical problems, the use of artificial intelligence, the improvement in the interaction and visualization of data and new implementations to improve the visual acquisition process using standard cameras in dynamic environments.
