Grant: Engage Plus Grants Program

The goal of this project is to enhance software that animates and visualizes volumetric data which is representative of granular substances such as soil and coal. Our overall approach will be similar to that used in our recent project with Serious Labs (Engage Grant). This approach emphasizes a three-dimensional (3D) grid of density values that indicate the amount of material in each part of the scene. With this volumetric approach, the 3D surface can be readily deformed to allow overhangs, caves, and tunnels. The volumetric approach is based on equal-sized 3D elements known as voxels, each of which occupies a small, unique, 3D subspace. Each voxel has an associated mass (or equivalently density, because voxels are equal-sized), which defines the amount of soil or coal occupying its space. To allow these granular substances to move in a physically realistic manner through space, the densities of the voxels will be changed as the software runs. Four features will be added to our approach: slippage, friction, transformation, and voxelization. First, a soil slippage model will be used to simulate the gradual movement of soil down piles and hillsides. We will enhance the soil slippage model to provide faster slipping than available with our current approach while maintaining a fast frame rate. Secondly, a friction model will be used to simulate the motion of the granular substance when it is on top or beside moving equipment. To do so, we will incorporate a voxel-based model of friction between external objects and granular substances. Thirdly, we will add the capability to transform chunks of solid terrain (coal or packed ground) into granular substances in response to impacts from external objects. Finally, we will enhance the speed and flexibility of incorporating complex geometric models, such as construction equipment, into the simulations. To achieve real-time performance, we will perform density movements in parallel by offloading them to the graphics processing unit (GPU) using a general-purpose GPU computing library, such as Microsoft's DirectCompute. We will implement our approach as a stand-alone C++ program.