Fast Collision Detection of Space-Time Correlation

To improve real-time performance and accuracy are key aspects of collision detection. In view of that conventional algorithms of collision detection spend a lot of detection time, this paper presents a advanced algorithm. We adapts a parallel method based on MPI. At the same time, we use temporal-spatial coherence and spatial subdivision algorithm. First, we subdivide the space into a series of voxels, and then we detect the state of the object. If the state is changed, we should build its list which is used to store its adjacent objects in voxel. We can begin with mark points. These mark points has independence, so the parallel method based on MPI can be used to speed up the collision detection. In a word, this algorithm reduces the times of collision detection and the traversing depth of the bounding box tree. The results of experiment prove that this method has real-time performance and superiority.