Fanuc 7th Axis with Irvision

    ok I have tested the "drifting" movement when I just move the RTU, it seems the error is slightly less (it drifts about 15mm instead of 20mm), but still far from perfect. Vision aside what else could cause this error? the robot is vision mastered (which I've heard is supposed to be more accurate than the factory mastering), and the robot is aligned on the RTU with an error of 0.0034 degrees.


    Since I aligned the robot on the RTU based on the world X motion of the robot, and it seems to be drifting in the X direction, maybe what PDL was saying about the gear ration could be an issue? it does seem dead on with just a simple tape measure measurement over 3 meters movement, and I figured since fanuc made it, it must be accurate :), but you never know.

    No I didn't redo the vision master, I'm under the impression that J1 and J6 is not effected by vision master, therefore I do not need to vision remaster. Am I wrong about that? because if I am, that could very well be my problem.

    so I just did a test with some disturbing results!


    I taught 4 points in a square with 1000mm between points. Measuring with a tape measure the distance between points measured from 980mm to 995mm, there was about 20 mm error in the X on one of the sides. So I think its obvious the mastering has to be out to lunch. I don't know how that can be since I vision mastered it.


    is there anything besides mastering that could cause these results?


    these tests were done with the RTU in the same position, so it should have no effect on this.

    I saved a point near the floor and made a mark with my frame teaching tool. then I moved that position plus 1000mm 3 times to make a square, then I measured between my marks with a tape measure. seems like a lot of error. The robot has j3 swung backwards for this test, because that's how it will be picking up the part in this case.


    what do you mean by enhanced master procedure? vision mastering?

    The floor is probably not perfectly flat, so that error will add to the robot. I have seen some pretty bad concrete floors. You would really need a very flat aluminum plate or a laser tracker to be sure that is all robot error.


    The enhanced mastering method is a procedure where you reach the same point with the arm flipped and wrist flipped. I'm sure it's on the forum somewhere. Some people have better luck with that method. I have seen mixed results.

    I have a crazy idea. :thinking_face:


    what if I run the vision mastering, and then run it again with the robot swung backwards, and take the average of those 2, since vision mastering is supposed to be very accurate.


    what do you think of that idea?

    Not a bad idea.


    So I assume this issue is causing your pick positions to be off, correct?


    It may be a simple solution to teach a different pick position for each robot snap position. I know it's not elegant but may be good enough to make up for inaccuracy.

    yes HawkME, The issue is exactly as you stated, and I could compensate for it with separate positions, but it would be nice to have it tuned properly because making new programs will happen regularly, and if I can fix it now, I do have the time.


    I like how you think pdl :smiling_face_with_sunglasses:


    I will give it a shot.

    ok so I did the vision mastering a couple times now, the first time I got corrections:

    J1 0

    J2 -0.426

    J3 0.268

    J4 -0.004

    J5 -0.006

    J6 0


    so I applied the new mastering and redid the whole procedure. the new values were:

    J1 0

    J2 0.000

    J3 0.001

    J4 0.001

    J5 -0.001

    J6 0


    so then I flipped the robot back around so its reaching behind itself, and this time the grid was on the floor instead of up on the 7th axis (which the robot hangs from), and I got the following corrections:

    J1 0

    J2 0.769

    J3 -0.401

    J4 -0.028

    J5 -0.102

    J6 0


    I am a little disappointed with how far out the numbers are with the last set of corrections, but that's how it is.

    So I'm thinking apply half that last correction and that's as close as I'll ever get it. does that make sense?

    Yes, that is basically what the enhanced mastering procedure is except using a pointer.


    On a side note:

    Mastering doesn't really make a robot accurate per se, it really just centers the error. The only way to completely remove the error is iRcal signature calibration.

Advertising from our partners