Closed mahaarbo closed 1 year ago
After some testing, the problem comes from replaceSubModel
not causing the collision geometry, the collision object or the broad phase collision manager to recompute the AABBs involved. A workaround is to call computeLocalAABB
on the collision geometry, and computeAABB
on the collision object, and then update
on the relevant broadphase manager after replacing the submodel.
This is an edge use-case that incurs extra computations for the most common use-cases if added, so I'll just close this issue.
Hi!
I'm testing out collision checking for a deformable mesh, and have run into a strange problem. If I define two collision objects, one is a box as an
fcl::Box
at a fixed location, and the other is aBVHModel<OBBRSSd>
representing a box that stretches usingreplaceSubModel
to update the location of the vertices, then I can test for collisions correctly using the narrowphasecollide
function. If I try to use a BroadPhaseCollisionManager where I have registered the mesh collision object however, it does not report the collisions correctly.Is this a known problem?
Code used to test this: broadphase_mesh_problem.cpp
Compiled with fcl version 0.7.0.