While going through the following chapter of gpu gems 3http://http.developer.nvidia.com/GPUGem ... _ch29.html
I came to the following excerpt from the chapter:
In our method, collision detection is based not on the polygons that represent the rigid bodies, but on particles, as done by Bell et al. (2005) and Tanaka et al. (2006). A rigid body is represented by a set of particles that are spheres of identical size
According to their description, they are doing the collision detection check with sphere particles instead of the whole rigid body which is itself is made of particles.
Looking forward to more on this issue.