About self-collision checking and inverse kinematics

edit

Hi there!

I am currently working on a teleoperation pipeline (more info: LINK) and just added the feature of self-collision checking. The process goes as follows: I first do the IK calculations, then check the result, if it would lead to self-collision, and if not so, the solution (desired joint positions) gets sent to the joint position controller.

In contrast to my solution, the get_constraint_aware_ik service of pr2_arm_kinematics_constraint_aware package seems far more complex. Thus, I am thinking, if I might have missed something important.

Could somebody give me a quick insight, if there are differences between both methods and if so, which they are?

For example, does the get_constraint_aware_ik service return a collision-free solution, which is closest to the optimal one, which however would result in collision? My design would just block any request, which would lead to self-collision and hence could result in positions, which are not the possible closest ones (only the last valid one).

I would already be happy with few lines about the general key differences!

Thanks for your help!

:-) Marcus