An Adaptive Bounding Object Based Algorithm for Efficient and Precise Collision Detection of Csg-represented Virtual Objects

Collision detection is the foundation of interaction in virtual reality. In Virtual design workspace, virtual assembly workspace, and virtual training, for example, the motion of virtual objects must be constrained by collisions with other objects to provide the sense of reality. Important human-object interactions such as grasping, hitting, moving, etc., in virtual reality (VR) environments are enabled and triggered by collision. A fast and precise algorithm for detecting the collision of virtual objects is therefore a critical component of VR systems. In this paper, an efficient and precise collision detection algorithm for CSG-represented objects in a virtual environment is proposed. In the pre-processing stage of the proposed method, CSG object models are converted into Brep object models and then both models are integrated into a CSG/Brep hybrid model by linking primitives in each CSG tree with corresponding faces in Brep. In addition, for each non-leaf node in the CSG tree, a bounding volume list is generated from the integrated CSG/Brep object model and attached to the nonleaf node. In the detecting stage, fast localization of possible collision regions is enabled by using a "divide-and-conquer" paradigm and an adaptive bounding volume selection strategy. Experimental results have shown the effectiveness of the proposed methodology.