Effective Octree Generation of Parts for Virtual Prototyping

Abstract

Abstract Virtual environment has become a future trend in design and development of new products, due to the current advances in simulation, computer graphics, robotics, and other related technologies. In order for the simulation engine to be efficient in performing tasks such as collision detection, the representation of the parts and components is very important. This paper presents an efficient algorithm to convert boundary representation (B-rep) to octree representation for efficient collision and interference detection. Traditionally, octree generation can be achieved through its original definition: build from the top level. This paper presents an innovative approach to build octree from the bottom level. This algorithm first transforms the boundary representation model into a triangle facet model and then a ray tracing technique is applied to scan all facets to create and record octants to a bit representation at the bottom level resolution. To save memory space, only the octants on the surface level are stored as black (solid) nodes. Finally a pointer representation of octree is built via the bit representation. The generation of a nine-level octree takes less then 60 seconds on a SunSparc 10. With this process, an octree can be quickly built for applications of collision and interference detection, such as robotic path planning, assembly planning, flight simulation, etc.