We advise running infield verification once or twice annually as a preventive measure. It is also good to monitor for picking accuracy or mispicks to check if picking performance deteriorates. Then, on such behavior, verify the camera for dimension trueness and correct for it if necessary. Infield correction was mentioned in the previous section. To read more about it, go to our Infield Correction page. Infield correction might require repeating the hand-eye calibration; see below.
If the picking accuracy starts to deteriorate, re-running hand-eye calibration might be necessary, in addition to infield verification and correction. Hand-Eye calibration was briefly explained in the previous section. For more details, go to Hand-Eye Calibration page.
Permanently mounting a Zivid calibration board in the scene allows for verifying camera dimension trueness as an automatic procedure, and without stopping the production. The same holds for performing hand-eye calibration; it can be done automatically, and without any mechanical changes to the cell.
Developing a system that allows quick mounting and dismounting of the Zivid calibration board on and from the robot is worthwhile. While the mounting process has to be manual, it is possible to automate the infield verification and correction process that follows. The hand-eye calibration process can also be automated.
Let us assume that the infield verification gives good results and you have performed hand-eye calibration the same way as in pre-production preparation. If you are getting bad hand-eye calibration residuals or if the residuals are good but the robot positioning accuracy is not, there might be something wrong with your robot. In that case, we advise contacting your robot supplier for more information about performing robot kinematic calibration and robot mastering/zeroing.
If external ambient lights have been modified you might want to redo color balance.