COLLISION DETECTION FOR CLOTH SIMULATION USING BOUNDING SPHERE HIERARCHY

Abstract

For the past 2-decades, the challenges of collision detection on cloth simulation have attracted numerous researchers.  Simple mass spring model is used to model the cloth where the movement of the particles within the cloth was controlled by applying the Newton’s second law. After the modeling stage, implementation of the collision detection algorithm took place on cloth has been done. The collision detection technique used is bounding sphere hierarchy. Then, quad tree is being used to partitioning the bounding sphere and the collision search was based on the top-down approach. A prototype of the collision detection system is developed on cloth simulation and several experiments were conducted. Time taken for this system to be executed is around 235.258 milliseconds. Then the frame rate is at the average of 22 frames per second which is close to the real time system. Times taken for the collision detection system travels from root to nodes were 23 seconds. As a conclusion, the computational cost for bounding sphere hierarchy is much higher because the bounding sphere required more vertices for generation process, however the execution time for bounding sphere hierarchy is faster than the AABB hierarchy.