Dynamic Collision Detection Using Space Partitioning

Abstract

A technique termed space partitioning is employed which dramatically reduces the computation time required to detect dynamic collision during computer simulation. The simulated environment is composed of two nonconvex polyhedra traversing two general six-degree-of-freedom trajectories. This space partitioning technique reduces collision detection time by subdividing the space containing a given object into a set of partitions. Using these partitions, all testing can be confined to the local region of overlap between the two objects. Further, all entities contained in the partitions inside the region of overlap are ordered based on their respective minimums and maximums to further reduce testing.