Real-Time Physics Simulation Forum

Computing collision radius of OBBs

2005-11-10T14:56:51+00:00

I believe I have resolved my own question, but just in case I'm wrong...

I need a 'collision radius' to calculate the magnitude of the impulse force from the collision (Baraff Siggraph course notes approach). A collision radius would be the vector from the center of mass of a rigid body to the point of the collision. I have been using the separating axis approach to detecting collisions between two OBBs during the narrow phase of collision detection and thought it could be used to figure out the collision radius.

I read and reread Eberly's chapter in his 3D Engine book and I think I understand the approach to finding the intersection point of two OBBs. The intersection point can then be used to create the collision radii and then the impulse magnitude. The Eberly book is the only reference I have found that describes how to find the intersection of two OBBs and is quite intense. I was seeking another reference to help me understand (even source code). I have looked through the source of several open source physics engines and have not found an example.

Is this a reasonable approach to calculating the impulse magnitude?

Mike

Statistics: Posted by mnjacobs — Thu Nov 10, 2005 2:56 pm

Computing collision radius of OBBs

2005-11-10T06:07:48+00:00

need to calculate the collision radii of the two colliding OBBs

Could you provide some more information? What is this 'radii of two colliding OBBs' ? And what do you need this for?

If you need a bounding sphere radius for a single OBB, you can take the length of the half-extents.

Erwin

Statistics: Posted by Erwin Coumans — Thu Nov 10, 2005 6:07 am

Computing collision radius of OBBs

2005-11-05T21:30:28+00:00

I am building a rigid body simulator using OBBs as the collision bounds of the objects. I have collision detection working via a bounding volume hierarchy of AABBs. I use the separating axis approach to detect the intersection of two OBBs and this appears to work. I have used a time interval bisection approach to reposition the objects.

I plan to use an impulse based collision response and need to calculate the collision radii of the two colliding OBBs. I have tried to understand David Eberly's works on the topic, but I am still confused.

Can someone provide an overview of how to find the collision radii? It seems to me that the separating axis approach could be used, but I can't figure out the next step.

Thanks for any help,
Mike

Statistics: Posted by mnjacobs — Sat Nov 05, 2005 9:30 pm