[ODE] Boolean operation on collision geometry

[Gary R. Van Sickle]

> Thanks Adam.
> 
> Noticed that Christian Larsen in
> http://q12.org/pipermail/ode/2004-October/014181.html talked 
> about a subtraction of a sphere from a box. I wonder if 
> anyone has done this work (or similar) or not (other than of 
> triangulation), as this particular case is just what I would 
> like to achieve.
> 
> Thanks
> 
> Gao Yang
> 

FWIW, I tried for a while to find something exactly like you're talking
about, but came up empty.  If papers addressing such an issue exist, I
couldn't find them on the web, though I would sure like to see them, because
it has the potential to make collision detection faster by (I suspect)
several O()rders of magnitude ;-).  What I finally decided to do is convex
polyhedral decomposition and then colliding the convex polyhedra, since I
have to do the decomposition for other reasons anyway.  But this ain't a
walk in the park either.

-- 
Gary R. Van Sickle
 
> 
> 
> Adam D. Moss wrote:
> 
> > Gao Yang wrote:
> >
> >> dGeomTransform is able to composite multiple collision 
> geometry into 
> >> one. Here's my question: If the composition in creating 
> >> dGeomTransform we call it an addition operation, is there 
> any way to 
> >> apply subtraction operation on collision geometry in forming a 
> >> composite geom?
> >
> >
> > Very difficult - check the list archives (keyword probably 'CSG').
> > Not impossible, but (I think the conclusion was) it would require 
> > being able to derive the intersection surface patches for 
> all of the 
> > supported ODE primitive collision combinations, and 
> inside/outside CSG 
> > between the resulting solids formed by those patches too.  No quick 
> > hackaround.
> >
> > --Adam
> 
>