iRVision Accuracy

  • Hi everyone,


    I was hoping someone might be able to help me improve the accuracy of a part picking program I'm using with an SC130EF2 camera and SR6iA/ R-30iB. Currently, when I go to pick up the part that I've created vision data for, the tool is a few millimeters off with no visible pattern (sometimes it is on the part, sometimes it is right next to the part, sometimes it is a little further). I was wondering if there was a way to take an image multiple times and then average the location - although I've been working with FANUC for a few weeks now I am still pretty new to how to program these types of things.


    I'm not sure if these details are important but I figured I'd add them:

    -When I set up camera calibration, the software thinks the camera is much further away from the grid than it really is. Hard coding this value does not improve accuracy.

    -When I teach the part in the middle of the tray I am using, anywhere outside of the bottom right quadrant of the tray (meaning the origin is in the top left) results in a limit error; I have to teach the part in the top left corner of the tray in order for the robot to reach anywhere on the tray

    -The part is seen as a circle from above so rotation should not really be a factor, although when the part is moved anywhere other than the middle of the tray (which is in line with the middle of the camera) the part is seen as a little more of an oval. It is still detected as a part either way.


    Sorry if any of this is confusing, please feel free to ask clarifying questions and as always any help is greatly appreciated.

  • I'm assuming a fixed mounted camera?


    How did you teach the grid user frame? If you used a pointer did you teach a good tool frame first?


    How did you set your Z height?


    How far away is the camera from the part?


    What Focal length lens are you using?

  • Good tool frame

    Good pointer tool frame for teaching user frame

    Good grid user frame

    Override focal distance (to lens size .ie 25mm)

    Application frame the same as grid frame used.

    Good application Z, measured with the robot pointer tool frame to the surface of image, robot in grid frame number used.

    AND! your GPM should only look at 1 plane of the part. Multiple height surfaces used in your GPM is a real bad no no.


    Any one of these bad, and you'll be fighting with it. Hope this helps

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account
Sign up for a new account in our community. It's easy!
Register a new account
Sign in
Already have an account? Sign in here.
Sign in Now