Skip to content

ENH/REFAC/BUG: CapsuleToCapsuleCD, Rigid Body & Collision Fixes

Andrew Wilson requested to merge andrew.wilson/iMSTK:CapsuleToCapsuleCD into master
  • Segment barycentric flipped #418 (closed)
  • Rbd stiffness is not actually stiffness, that was a misunderstanding. Its actually a "baumgarte stabilization term", which is not particularly intuitive, it can be used to vary stiffness, but should not be interpreted as stiffness.
  • Edge to Edge closest point computation had some niche cases I couldn't quite quantify/overcome. I replaced it with the one from the orange book which works well.
  • The contact points on the rigid body were slightly off in the case of PointDirection. This is only used in the constraint for computing r, moment, for which an angular velocity is produced, so it was basically unnoticeable, especially under small rotations. It was using pt + depth * dir for the contact point on the body. Which assumes pt is the deep point in the body, for which to resolve. This was not the case in the CD methods. CD methods report the contact point for/ON the body.
  • Adds CapsuleToCapsuleCD, which finds the closest point between two edges then performs sphere CD.
  • Adds some initial visual testing for RigidObjects.
  • Adds an example for capsule to capsule collision for which two laparoscopic tool capsules can collide.
Edited by Andrew Wilson

Merge request reports