Real-Time Physics Simulation Forum

Try to clear the all manifolds related to your ghost object at each step

yes

mmmm my problem is still there. I must move boxes, and a capsule will be an approximation too loose. I must keep boxes.

Sorry, I don't know: maybe the btInternalEdgeUtility could solve the problem, but not in my case. http://www.bulletphysics.org/Bullet/phpBB3/viewtopic.php?f=9&t=4603 I can just warn you that cp.m_lateralFrictionInitialized = true has changed in cp.m_contactPointFlags |= BT_CONTACT_FLAG_LATERAL_F...

Hi,

look here
http://www.bulletphysics.org/Bullet/php ... 2&start=30
In short: fix the normals and call setContactProcessingThreshold(0) on the object (I don't know if both the ground and the moving object).

Hi,

I'm exactly in your situation. How have you solved the problem?