[ODE] Russ' plans for ODE?

GARY VANSICKLE g.r.vansickle at worldnet.att.net
Wed Apr 21 02:25:30 MST 2004


> Features I would personally be interested in:
> 
> Buoyancy (see
> http://www.cambrianlabs.com/Mattias/DelphiODE/BuoyancyParticles.asp for an
> implementation)
> Cloth simulation & interaction with geoms
> Convex hull collider
> 
> What would everybody else like to see?

Ah, a wishlist!  I'm way better at writing them than writing code ;-):

- Collision Detection, not Interference Detection
 What most everybody (here and pretty much everywhere else) calls "collision
detection" is really only "interference detection".  These are not the same
thing, as any fast-moving small body passing right through an intervening
thin body can attest.  I think the "witness plane" method can do this fairly
cheaply for convex polytopes (and should also be applicable to the smooth
geometries already supported in ODE as well).  At a minimum this could
prevent situations where say a box falls slightly more than halfway through
a thin plate, and is then incorrectly pushed through to the other side by
the reaction force.

- "Penetration-Free" Simulation
 Given an initial state wherein no bodies intersect, at any and all future
time steps there are still no intersecting bodies regardless of forces and
torques.  Whether there is a known solution to the rigid body problem with
this constraint I know not.

- Rolling Friction (I think somebody already said that).
 This seems like it would need to be "special cased"; is that correct?  That
would be unfortunate.

- 100% Foolproof Automatic Object Deactivation/Activation
 Again, this might still be a thesis or two away from being possible, but
would be hella-cool.

...that's all for now, time to sleep.  Basically I want everything ;-).




More information about the ODE mailing list