Minimizing error values with iRVision

    I am currently in the process of calibrating 8 cameras with 2 robots using the Robot-Generated Grid Cal. Tool. Each robot has 4 cameras associated with it and they are all set up in a horizontal line. I have an extension tool on my EOAT so the robots can reach all of the cameras. I have got them all calibrated with decent error values. My minimum error values are all under 1. My maximum error value differs a little more which is discerning. The maximum error ranges from around .8-2 on Robot 1, and around 1-2.1 on Robot 2. I'd like to get the maximum error value on all of my calibrations to below 1 if possible. Any tips on how to do so would be greatly appreciated. Thank you.

    You can delete the outlier data points in the cal grid data. That should reduce the error somewhat.

    Also, make sure you are running the IRVBKLSH karel program before each calibration point (if you are not using the autocal utility, it does this automatically.).

    Check out the Fanuc position converter I wrote here! Now open source!

    Check out my example Fanuc Ethernet/IP Explicit Messaging program here!

    I assume you are asking about the IRVBKLSH program.

    IRVBKLSH cancels the backlash in the robot's arm gear train by preforming a small move. Once the backlash has been canceled, the arm's real position better represents the calculated position.

    Check out the Fanuc position converter I wrote here! Now open source!

    Check out my example Fanuc Ethernet/IP Explicit Messaging program here!

    I am using autocal & it runs the IRVBKLSH program. My max error value isn't horrible, I think my worst is 2.1pix. But I'd like to get them all closer to 0. I was just curious if maybe anybody knew of any ways to reduce this. I have tried a few things like messing with the exposure of changing the intital position around. Nothing seems to really get it lowered though. I am starting to think maybe it is a little too shaky on some of the far points since the target is on a 3 foot pole extended off of the EOAT.

    Edited once, last by romanzo (July 25, 2017 at 5:16 PM).

Advertising from our partners