Is anyone having an issue getting the camera calibrated?

Every time the calibration process stops before taking the 18 calibration poses.  I have tried different backgrounds and different starting positions (making sure to update the "snapshot pose" feature).  When I press "calibrate," the robot moves to a few locations taking pictures (differs in number depending on where I start relative to the board) and then for a few seconds the picture area on the left shows the Poses 1-9 grid.  After that the camera begins flashing on and off and the "Calibrate" button is available again.

I know in the training something that was discussed was having error messages, or failure notices since those are not explicit.

Any advice, thoughts?  I may be missing something very simple.

I would check if you don't have any shadow on your calibration grid especially the gripper's shadow.

Other than that, @PierreOlivier_Proulx and @MarcAntoine_Lacasse, do you have any ideas?

@Catherine_Bernier  in this case I do not have a gripper on the camera, just doing a part test.

@PierreOlivier_Proulx @MarcAntoine_Lacasse   any advice would be greatly appreciated.

@Grady_Turner Check in the Polyscope log what it says. Can you take a picture and post it. I guess after the robot has moved to its first calibration pose, the grid was lost probably due to a focus issue or a reflect on the grid. Can you also take a picture of the wizard once the Calibrate button has been reset. This will show us what the camera is looking at when the problem occurred.

Regarding adding explicit error messages, it has been integrated. I could probably send you an update tomorrow.

@PierreOlivier_Proulx   Thanks for the information about the log file, I did not know the routines wrote explicitly to the log.

It looks like every time the system calculated the 18 poses, pose 11 "inverse kinematic" request returned false, and so the calibration failed.  This was consistent on pose 11 every time.

I am doing this on a UR3, and as I mentioned, I tried several starting locations knowing how the robot might move, but my guess is that pose 11 is either too close to the base or too far away from the base.  Any advice?

@Grady_Turner With the UR3 you need a different grid because of the reach of the robot. With the big grid, after the centering step, the robot is probably already too extended to be able to reach the calibration poses.

With the UR3, the squares of the grid must be 8.24 mm. You will find attached the original grid. To achieve the right dimensions, I printed it with a 58.8% scale factor. You should still verify the dimensions after printing.

@PierreOlivier_Proulx  Do the images need to be in color? I know the camera does not see color, but it could interpret the colors differently than if it sees actual gray scale of those colors.

@Grady_Turner You mean the colored circles on the calibration grid? Their color does not matter. In fact I think there is no use yet for those circles.

@Grady_Turner Hi Grady, did you manage to have a demo unit running on the UR3?

@PierreOlivier_Proulx  Yes I got it working with the 58.8% scale calibration board.  Thanks!

@Grady_Turner Have you been able to teach a part and pick it with a decent grasp position relative to the part?

@PierreOlivier_Proulx no not yet, the testing I did was just to see if we could get a good model of a part (it is a very reflective hollow cylinder).  I did no testing with trying to pick the part.

@PierreOlivier_Proulx
UR3 with Vision. We use your CalibrationBoard!. 

https://youtu.be/NLip_5yUoCY