A GPU-based integrated approach to simulation for deformable surface meshes

Surgical techniques have evolved from direct hands-on maneuvers to indirect minimally-invasive procedures, learning which involves using virtual simulation environments with standardized exercises typically comprising scene representation, collision-detection, force feedback, and rendering. Key challenges are the need to improve speed and realism of the simulation while being run on consumer-grade computing platforms. This thesis aims at overcoming these challenges by developing an algorithm to utilize the Graphics Processing Unit (GPU) as a parallel processor. The approach presented consists of three phases: Off-line surface wrapping, implicit integration algorithm for deformation, and tactile feedback. A prototype implementation using the NVIDIA GeForce 7600GS is presented. Interactive surface wrapping models a cloth-like surface into a closed mesh while the deformation phase is a GPU-based parallelized Implicit Euler method. Point-based haptic interaction with virtual coupling provides tactile feedback. The results show significant speedups, a maximum of 6.5 times, for the GPU-based simulation over its CPU counterpart.