InitRech 2015/2016, sujet 10

SOFA: A Multimodel Framework for Interactive Physical Simulation

Summary

Interactive simulations of objects need multi-disciplinary expertise and from the software engineer's point of view, efficiency is key to obtain a real-time simulation. In order to help experts design such simulations, the SOFA Framework is a strong tool that uses several numerical models to compute data and offer the best performances. In addition, SOFA allows collaboration between specialist from various domain, using encapsulation of the simulation components to design them independently.

This article describes how models are processed and explains the latest upgrade to the framework.

Commonly, 3 models are used : A deformation model, which describes the system. A collision model, which computes how it will react to the environment using the data from the previous model, and updates it. In fine, a visual model to display with best accuracy the computed simulation.

Using this method, the framework is allowed to efficiently compute data using Fine Elements Methods, namely triangle and tetrahedra, give real-time response, and render a simulation accurate enough for humans to use it.

Main Contribution

Numerous numeric methods have been added to boost the software's efficiency. Using one or another method depends on the situation and the needed accuracy and speed. Namely, collision model, data structures, . Conjugate Gradient, Also, they added GPU support, that eases the computer process the data.

In order to get real-life images for simulation, they implemented their own rendering library based on OpenGL, for the Visual Model. This library can be used to render visual effects such as X-ray and fluoroscopic images, which can be used for radiology simulations.

Non mechanical phenomenon such as electrical waves (e.g. for cardiac electrophysiology) have been implemented.

Applications

The obvious application domain is the medical simulation. A surgeon can practice, using an accurate model to gain experience without risking anyone's life. They aimed SOFA for adaptability and performance, making it a precise and professional tool to learn.