Differences between V-Collide and prior work

I_COLLIDE and V-Collide

1. I_COLLIDE requires that models are decomposed into (unions of) convex polytopes. V-Collide can handle any polygonal model, without requiring topological cleanliness.
2. V-Collide does not contain any I/O routines.
3. V-Collide requires that input polygons be triangulated.
4. I_COLLIDE reports the distance between objects, but V-Collide only reports when pairs of objects collide.
5. V-COLLIDE supports dynamic addition or deletion of objects from the environment.

RAPID and V-Collide

1. V-Collide keeps track of where objects are, so that if objects do not move between queries their locations need not be resupplied to the collision detection system.
2. V-Collide supports many simultaneous objects, where RAPID only handles two.
3. RAPID can report exactly which pairs of triangles collide, but V-Collide only reports to object precision.