Kuka robot doesnt repeat one specific position?

    I have a big robot (KR 210 R3100-2 / KRC4 8.6.11) with a 202 kg gripper.
    I ran the kuka load data determination program with this robot. I did the mastering with kuka EMD.

    But I'm having trouble with a specific position. When the robot reaches the final position to drop the parts, sometimes the position is different. In the smartPAD the position is always the same, but physically the position is a bit different (I have 4 locators in the corners of the gripper)

    It is a linear point, at 0.2m/s and it is NOT continuous, my position shouldn't vary, right?

    This is my point:

    ;FOLD LIN pDropPart_IMM_prueba Vel=0.2 m/s CPDAT89 Tool[3]:tGripper3 Base[3]:bIMM ;%{PE}
    ;FOLD Parameters ;%{h}
    ;Params IlfProvider=kukaroboter.basistech.inlineforms.movement.old; Kuka.IsGlobalPoint=False; Kuka.PointName=pDropPart_IMM_prueba; Kuka.BlendingEnabled=False; Kuka.MoveDataName=CPDAT89; Kuka.VelocityPath=0.2; Kuka.CurrentCDSetIndex=0; Kuka.MovementParameterFieldEnabled=True; IlfCommand=LIN
    ;ENDFOLD
    $BWDSTART = FALSE
    LDAT_ACT = LCPDAT89
    FDAT_ACT = FpDropPart_IMM_prueba
    BAS(#CP_PARAMS, 0.2)
    SET_CD_PARAMS (0)
    LIN XpDropPart_IMM_prueba
    ;ENDFOLD

    DECL LDAT LCPDAT89={VEL 0.200000,ACC 100.000,APO_DIST 0.0,APO_FAC 50.0000,AXIS_VEL 100.000,AXIS_ACC 100.000,ORI_TYP #VAR,CIRC_TYP #BASE,JERK_FAC 50.0000,GEAR_JERK 100.000,EXAX_IGN 0,CB {AUX_PT {ORI #CONSIDER,E1 #CONSIDER,E2 #CONSIDER,E3 #CONSIDER,E4 #CONSIDER,E5 #CONSIDER,E6 #CONSIDER},TARGET_PT {ORI #INTERPOLATE,E1 #INTERPOLATE,E2 #INTERPOLATE,E3 #INTERPOLATE,E4 #INTERPOLATE,E5 #INTERPOLATE,E6 #INTERPOLATE}}}

    Any ideas on how to improve repeatability? Thanks!!!!

    DOjYtvX.jpeg

    Edited 4 times, last by Leonel97 (May 23, 2024 at 8:22 PM).

    2 things come to mind for me:

    1. Where the robot sits and where the part sits; are those modular pieces? Is it possible that the concrete beneath your robot is shifting this spring?


    2. If you are using load data determination, do you have a load cell on there? If so

    - You could use the load cell to locate the part. If you have access to the RSI and Force Torque Control add on's.

    - When you do the load determination, do you account for the load cell? ir is it a value after the load cell?

    How much is the repeatability error?

    Is the robot approaching this point from the same previous point every time? If not, this can cause backlash effects to play into your repeatability.

    Are the Base or Tool values ever adjusted by the program?

    Is there anything between the robot and the tool, or between the robot and the fixture, that has any possibility of mechanical slippage, flex, etc?

    Can you identify anything that's not the same between the times the move is repeatable and not? Something heavy moving nearby (I've seen concrete floors flex in ways you would not believe), heat/cold (sometimes thermal expansion can surprise you), more/less weight being carried by the robot, etc.

    Quote from work_BR

    2 things come to mind for me:

    1. Where the robot sits and where the part sits; are those modular pieces? Is it possible that the concrete beneath your robot is shifting this spring?


    2. If you are using load data determination, do you have a load cell on there? If so

    - You could use the load cell to locate the part. If you have access to the RSI and Force Torque Control add on's.

    - When you do the load determination, do you account for the load cell? ir is it a value after the load cell?

    1.- Im not working wit modular pieces.

    2.- I didn't use a load cell, should I use one? I only ran the LDD program.

    Quote from SkyeFire

    How much is the repeatability error?

    Is the robot approaching this point from the same previous point every time? If not, this can cause backlash effects to play into your repeatability.

    Are the Base or Tool values ever adjusted by the program?

    Is there anything between the robot and the tool, or between the robot and the fixture, that has any possibility of mechanical slippage, flex, etc?

    Can you identify anything that's not the same between the times the move is repeatable and not? Something heavy moving nearby (I've seen concrete floors flex in ways you would not believe), heat/cold (sometimes thermal expansion can surprise you), more/less weight being carried by the robot, etc.

    If I leave the gripper inside the mold (as the OP picture) I don't have any variation, I can leave the robot there and the next day is on the same position. And I can run cycles without any problems. But if I leave the robot in home position for a while, then I go to the mold position, the locators of the gripper doesn't fit in the locators of the mold (home position is the typical position for home and the gripper is in vertical). By 5mm aprox. Sometimes 5mm up and sometime 5mm down. That's a lot.

    I have an approach point 1m from the end point (the mold):
    LIN pDropPart_IMM_Aux2 CONT Vel=1 m/s CPDAT80 Tool[3]:tGripper3 Base[3]:bIMM ;%{PE}

    then, i have another approach point 10cm from the end point:

    LIN pDropPart_IMM Vel=0.5 m/s CPDAT91 Tool[3]:tGripper3 Base[3]:bIMM ;%{PE}

    then, my final point

    LIN pDropPart_IMM_prueba Vel=0.2 m/s CPDAT89 Tool[3]:tGripper3 Base[3]:bIMM ;%{PE}

    those 3 points are a lineal movement.

    I would check the gripper, make marks on the gripper to see if something on the gripper changes also, I had bigger robots with bigger tools and the robot was ok.

    Usually when I observed the programmed points shifting was due to mechanical issues eihter on the gripper or something else in the cell moved or wasnt bolted to the ground, etc...

    need to get numbers... only when you keep track of deviations (direction, amount) you can determine if this is really random or not. making sure all anchor bolts/tooling have paint marks, checking backlash, etc are basic steps.

    next check the program... any interrupts? anything using RESUME instruction? any calculation/modification of tool or base or load data?

    or just create small program with just two points, one of them being the same as problem point. no tool or base manipulation. then run them in a loop... do you get any randomness in this case too? if so, something is mechanically wrong. if no, problem is in the program.

    1) read pinned topic: READ FIRST...

    2) if you have an issue with robot, post question in the correct forum section... do NOT contact me directly

    3) read 1 and 2

    also check if the "drift" occurs if robot is interrupted by operator entering cell etc.

    and check KSS version and ask KUKA for support. there may be a bug you are not aware of.

    for example just recently i had to upgrade KSS because robot was misbehaving - after running for while, if program was stopped, and resumed, it would select nullframe as a base. even if manually selecting instruction and trying to perform BCO everything would be fine until Start is pressed... at that moment base was changing to 0 and robot was en route to a different dimension. KSS upgrade fixed this and few other smaller things. with each generation of KSS it takes fewer releases to reach some stable level, but every time major version changes, things are started from scratch (or close to) so new releases are always more suspect...

    1) read pinned topic: READ FIRST...

    2) if you have an issue with robot, post question in the correct forum section... do NOT contact me directly

    3) read 1 and 2

    I haven't had any problems since last Thursday (that is a lot of time)
    And the only thing I did was change this point

    I have an approach point 1m from the end point (the mold):
    LIN pDropPart_IMM_Aux2 CONT Speed=1 m/s CPDAT80 Tool[3]:tGripper3 Base[3]:bIMM ;%{PE}

    Before the point was with Base[0]
    And the day I made this thread I changed it to Base[3] so that it had the same Base as the other two points.
    I don't know how a BaseFrame corrects this problem because that point it is an approach point 1m from the end point, and i touchup the point in the same position but with Base[3]... Or maybe the mechanical guys adjusted something without me realizing it and didn't tell me about it?

Advertising from our partners