MattFanguy
You could program multiple models, and have a different color constraint for each.
For example, you would have two different "Camera Locate" Nodes in your program each trained to look for the same object, but with different color constraints. Your program might would start by taking a snapshot and detecting only the green colored objects of interest. If it sees yellow, it will ignore them no matter how confident it is in their shape recognition. You would then have a pick and place process within that Camera Locate node to flip over the green ones, place them where they need to go, and then return to the snapshot position when it has looped through all the green detected objects.
After that, you'd have your second Camera Locate node directly after the first one to take another snapshot, detect the yellow ones, and repeat the pick and place process this time without flipping them.
bjoern
Dear all,
I have to program a pick and place application. The robot needs to detect objects of size 50mm x 3mm. On one side they are white and on the other side they are green on the bottom and yellow on the top. We need to pick the objects and place them into a tray with the yellow side to the top and the green side on the bottom. The objects are made available to the robot on a vibrating plate.
Hardware setup: UR5e with a wrist camera and a customized ePick solution.
The detection of the objects is working well on a white background. But I cannot control the orientation of the parts. Sometime the yellow side is on top and sometimes the green side.
Any suggestions on how this can be solved?
Thanks a lot,
Björn