Hi,
This question arises from my recent post in the forum: https://www.robot-forum.com/robotforum/kuk…/robot-way-off/
Thanks for all help, now the robot (a Kuka KR5) is working properly.
But from the mastering experience, it looks like an exact mastering is a must for the assurance of the position accuracy (not the repeatability).
We are using a self developed 3D camera and use it as an optical TCP. We developed some algorithm to automatically perform a 8 point calibration (similar to the 4 point TCP calibration), but we never get any result less than 3mm error.
The way I look at it, mastering an axis could easily be off by some angle (or equivalent to maybe 0.05mm travel in the circumferential direction when the dial gage meets the notch pad). This could set the position at the end to be off by a few mm.
So, I suspect, typically, the robot, in the absolute accuracy sense, could be off by a few mm in the full work range. Is this what expected?
If this is the case, we will not be able to successfully using the 4 or 8 point TCP program to calculate the TCP point to accuracy better than a few mm because the mathematical model is not accurate to a few mm.
But now I have an idea.
Can we directly using the DH parameters and using the axial data for the six axes to not only calculate the TCP but also the mastering position of each axis? Of course, we will need to have the official DH parameters for my robot. Not sure if this is possible.
Any thoughts?
Thanks,
Jay