Background for Camera

Tyler_Berryman

We got a question from @nbouhet during the vision system training :

What is the impact if your background becomes dirty in the ability of the Camera to detect objects?

Etienne_Samson

@Tyler_Berryman @nbouhet release 1.0.0 of the URCaps now has the Detection Threshold feature (this was not available during the training under beta 0.9.0). You should adjust this threshold so that you can still have a reliable object localization when your background gets dirty. Of course at some point you should have maintenance intervals where you clean that background.

Some general hints: if your threshold is very low, you will probably get false detection when your background changes. You should get the highest possible score where your part is always detected. If your parts change a bit (the part themselves could be dirty), you should adjust to a lower threshold. There is a sweet-spot to find when trying your part during the Object Teaching last step, Test & Save.

You can find more information on the Detection score here: http://support.robotiq.com/pages/viewpage.action?pageId=10847510