Towards time-critical collision detection for deformable objects based on reduced models

Abstract

Topics of physically based modeling of deformable objects and collision detection have been extensively researched. Nowadays, the combination of GPU techniques and multiresolution physical models allows interactive simulations of complex deformable objects with a large number of polygons. To achieve this, the geometry of the object is separated from the deformable model in order to represent the latter at different levels of resolutions (reduced models). Recently, James and Pai extended this idea to collision detection. They used the coordinates of the reduced model to update a sphere-tree hierarchy during object deformations. Moreover, many interactive simulations require a constant frame rate which can be obtained by using a time-critical approach. This approach allocates a given time-slice to the collision detection process, so it can be interrupted if the assigned time is expired. To the best of our knowledge, timecritical approaches have been only applied to rigid body collision detection. In this work, we propose an algorithm to carry out collision detection for deformable objects using an interruptible mechanism.