Vision Offset Relative to Userframe

    Hello

    I have a camera mounted on my EOAT. It provides offsets of my target object to the robot via plc. I have created a tcp to be aligned with my camera's coordinate system.

    I am having a hard time wrapping my head around how to apply the offsets provided by my camera so it is relative to my current user frame. I have the vision support tools option which has the Inverse and Matrix programs just not sure how to transform the offsets from the camera so it is relative to my user frame.

    Any help would be appreciated, thanks

  • HawkME August 29, 2023 at 3:59 AM

    Approved the thread.

    Hi,

    do you have Fanuc iRVision camera, or other brand?

    What values do you get, X,Y,rZ or more?

    If you have camera parallel with working TCP, then most time is enough use TOOL_OFFSET.

    You make picture, set it as zero value, then teach pick up possition. For next picture you store camera values in PR and use it for offset.

    If you get absolute position values, then you need to use (base) OFFSET.

    Also depends on the application, maybye you can tell us more details.

    Hello

    -I have a Cognex camera, no iRvision.

    -Application is only required to pass X,Y co-ordinates

    In my application I have a 5x5 rack, each position in the rack has 3 mating features for the 3 parts the robot is holding. I need to locate X,Y position of the feature and place each part. Once I place one of the parts I cannot move back to take another image because each part the robot is holding is connected together with a harness. So I need to take 3 images and get 3 offsets to place each part. I was thinking I would teach 3 positions in the first shelf of the rack and call this zero position, meaning all other positions would be an X,Y offset of the taught positions. I have UF[1] that lines up with the rack and have taught my "zero" positions in this frame. Correct me if I am wrong, the offsets I get from the camera will be relative to the tool frame at the position the image was taken. If I apply the offsets as tool offset to the "Zero" positions it would be relative to the orientation the zero positions were taught in and not to the camera position.

    I was hoping to be able to transform the camera offsets so it is relative to the UF[1] frame.

    I am rethinking solution because I forget to mention one step.

    You should have calibration grid for camera.

    Put it on rack and measure it as UFx, then you need to do camera calibration (set correct distance from grid and use this distance for all camera processes).

    Then you remove grid and do reference pictures (store it in 3 PR as you said) and teach reference positions in first rack.

    If you are sending position of robot to the camera system, then camera should return offset values relative to this UFx, you should use (base) OFFSET and have all movements in measured UFx.

    Then for other rack you use OFFSET in UFx for camera position and just OFFSET (camera values) for drop positions.

    If you are not sending position of robot, then camera have no information about its position, so you get values relative to camera position and you can use TOOL_OFFSET. And you don need to use measured UFx. For other rack positions you need to combine OFFSET in UF1 for camera position and OFFSET in UF1 + TOOL_OFFSET (camera values) for drop positions.

    by + i mean "L[x] ... OFFSET [PRx], TOOL_OFFSET [PRy]"

Advertising from our partners