Collision Detection Research Benchmarks / Evaluation

[gilbo]

Hi all,

Apologies up front for being impatient and just posting a question to the forum; I didn't spot a previous discussion on this topic from a quick scan of pinned topics.

I'm looking at doing a research project on a collision detection system, focusing more on collisions with lots of simple objects, cloth self-collision. (as opposed to the complex rigid polyhedra colliding case) In order to keep my projects focused, I prefer to start work on the evaluation part of the project first.

I'm trying to figure out candidates for two main dimensions of the evaluation:
1. state-of-the-art collision systems
2. appropriate benchmark test cases

I haven't been able to find any standard benchmarks available online, but maybe I've just been looking in the wrong places.

Bullet seems like an obvious choice for a state-of-the-art collision system, but maybe folks on here can think of other systems that are better in some cases/scenarios than Bullet, or are of comparable quality.

Again, I should stress that I'm asking from the perspective of doing a research project, so I'm not interested in starting any holy wars. I just want to make sure that we fairly and accurately represent the state of the art.

Thanks for any help you can offer,
-- Gilbert Bernstein