The Research Of Continuous Collision Detection

Collision detection, a critical role in the physical system, is widely used in many fields, such as computer graphics, robots, CAD/CAM and physical simulation etc. However, to create a physical sense of reality, all primitives are forbidden to penetrate with each other and all collision can be detected in these systems. As traditional discrete collision detection cannot meet the need of high precision, the continuous collision detection has deserved a great attention. In order to improve the efficiency of the continuous collision detection, the efficiency of different bounding volumes in the high-level culling is studied and some methods of low-level culling are proposed. Our research and innovation are described as follows:In this paper, the summary of collision detection algorithm is introduced and the application scenarios of the different collision detection algorithms are analyzes. The second chapter focuses on introducing different bounding volumes. After introducing different bounding volumes, the paper summarizes the advantages and disadvantages of the bounding volumes, so it can provide the basic theory of how to choose the bounding volume in different applications. In this paper the whole process of continuous collision detection algorithm based on bounding volumes is proposed. According to the comparison experiment of different bounding volumes using in the continuous collision detection algorithm, the most suitable bounding volume for complex deformable model is obtained.A continuous collision detection algorithm based on mesh topology structure is proposed. Two steps are used in this algorithm. Firstly, the adjacent triangles are divided into adjacent triangles by edge and adjacent triangles by vertex based on the mesh topology. The optimized adjacent triangles guarantee that not all9edge-edge tests and6vertex-face tests need to be executed. Secondly, before executing the accurate collision detection of the primitives, using additional bounding volumes for vertexes, edges and faces and the database of the elementary tests that have been executed to further remove disjoint primitives. The experimental results show that the proposed algorithm reduces the number of unnecessary elementary tests and improves the overall performance of continuous collision detection.A continuous collision detection algorithm based on one dimensional subspace filters is proposed. Each elementary test in3dimension space of primitives can be broken down into a coplanar test and an inside test. We further optimize the continuous collision detection based on the non-penetration filter that if the primitives are not coplanar during the time interval then they cannot collide. The non-penetration filter only removes primitives that are not coplanar at the whole time interval. We can further remove vertex-face tests that are coplanar at the time interval based on the one dimension subspace filter. The experimental results show that the proposed algorithm has higher culling efficiency on vertex-face tests compared with non-penetration filter.