Thanks to the debugging system from last post, I was able to pinpoint the bugs fairly quickly. However, as I tweak the algorithm to improve its numerical robustness, it becomes obvious that the problem is not as simple as just making the algorithm robust. In order to elaborate my concern and problem, here’s a review of the main algorithm.
Recap of Faurby’s
As I have stated in my previous posts, the current collision system is based on Faurby’s paper, which is really just a simple swept sphere vs triangle mesh collision detection & response routine. Here’s a pseudo-code version:
Read More →