I just encountered an issue with TCP accuracy of an invert mounted robot IRB1300 on a 3-axis gantry. My application is welding with laser seam tracking which requires a highly accurate TCP. When I defined a TCP, the mean error is around 0.5-0.7mm, but the actual deviation is much bigger when I tried to rotate the TCP around a spike (3-5mm). My robot is axis calibrated and ran by Omnicore. Do you guys have any suggestions or troubleshooting ideas? Thanks.
Hallo,
If you use 4-point + z method, the mean-error should be 0.1 -0.2 mm.
For any reason, one cannot achieve this for a robot.
Try to make correction manually:
Define TCP, position your tool above your tip, so that flange z is vertical. Turn around z in world-coodinates (do not turn around axis 6) ± 90 degrees.
The deviation shown is your deviation in x and y. Correct this values manually in your TCP, first x than y.
Turn the tool around world-y or x to correct the z-component of your TCP in the same way. (the difficult part).
Try in any case to make your TCP in a similar position to your welding/working position.
Do not use an overturn of axis 4/6 or a flip of 2/3.
Use, if possible a bulls-eye to make your TCP.
The bulls-eye routine will not calculate but measure your TCP.
If the latter is not available, find different TCP’s for different working positions.
(Welding tool in position fore-hand TCP1, tool overhead TCP2 etc.)
Finally, an absolute accuracy measurement of your robot will solve all problems.
Has the robot system been installed as inverted? Are you only using the robot axes when you position the points to define the TCP? Is the gantry rigid enough or is it flexing when the robot moves?