An improved method of continuous collision detection using ellipsoids

Continuous collision detection for virtual scene simulation is an important research field. In order to avoid penetration during the collision detection process and to ensure the completion of the collision response, the paper optimized the algorithm steps based on the traditional ellipsoid scanning method, and for the region based task allocation model of the physics engine, it gave a mechanism for the rapid index based on marked sub-regional. In addition, forecasting mechanism used during the process of collision detection effectively reduced the sum of collision detection operations within one simulation cycle. Experiments show that the algorithm optimization we proposed has significantly improved the efficiency of the collision detection, and the space division mechanism can be applied to parallel collision detection of large-scale and large-quantity objects in complex virtual scene applications.