Collision Detection for Arbitrary Polygonal Objects

V-COLLIDE is a collision detection library for large environments. It is designed to operate on large numbers of polygonal objects. It makes no assumptions about input structure and works on arbitrary models, also known as polygon soups. V-COLLIDE uses a three-stage collision detection architecture:

The n-body routine uses coherence between successive time steps of a simulation to perform well in animations and moving simulations. The hierarchical oriented bounding boxes (OBBs) and exact collision routines are taken from RAPID, a component of V-COLLIDE which is also available as a stand-alone package.

We maintain a list of differences between V-COLLIDE and RAPID and V-COLLIDE's predecessor software, I-COLLIDE. Click here to see a brief description of each of the systems, and for guidance on which may be best for your purposes.

Code Available

The latest V-COLLIDE source code, version 1.1, was released on February 1, 1998. Please use this request form to download the latest version. It contains source code for the collision detection libraries, brief documentation, and several sample applications, and is freely available for non-commercial use. If you put it to good use, let us know about it.

See also the release notes for the current version, previous and future versions.

V-COLLIDE Developers

Geometric Algorithms for Modeling, Motion, and Animation

Copyright © 1997-1998. Last modified: September 1, 1998.
Maintained by: geom@cs.unc.edu.