Thank you SkyeFire.
I double checked the tool and base defs and they are correct. i.e. they are the same (within the tolerence you mention) in both simulation and real world.
I was thinking of a couple things:
1. Is there a way for KUKA to take a Quaternion - which would remove the whole multiple solution issue?
2. I found this post: https://www.robot-forum.com/robotforum/kuk…rmation-matrix/ And I found the util, but my German is sketchy (not absent just sketchy). I havent figured out how to use it - but if I generate a rotation matrix using Q=rotz(gamma)*roty(beta)*rotx(alpha) - giving a 3x3 mx. Can I give this to KUKA and it sort out its own internal rotation?
Unfortunately extrinisc vs intrinsic is pretty NB for direction or application of rotation, but how that relates to KUKA I dont know. When KUKA applies a rotation to a tool does it do it with respect to its own frame or the frame of the world? (so if the part is sitting at x0y0z0a0b0c0 how does KUKA define where it needs to be wrt to the part)
You are right about "It doesnt work" So rather - it obtains the xyz position but the angle of orientation is incorrect, i.e. the tool orientation is wrong at that position, so rather than does not work it has a bug.