Is there a way to obtain the number of regions detected from an object with the wrist camera

Sergio_Rendon

Im working on an application that needs the detection of a part with a lot of holes in it, i use the camera locate node and the camera detects all the circles in the part is there a way to differentiate each part by the number of holes in it?

David_Gariepy

Hi @Sergio_Rendon,

The camera cannot count the number of object or features. Also, it is not possible to differentiate many objects with only 1 camera locate. You'll need a camera locate for each part.  If the parts are similar, you'll need to search for the part with the most feature first. Then if you don't find any, look for a part with less feature and so on.  If you could share some pictures of your parts we might be able to give you some tips. 

 

Sergio_Rendon

hi @David_Gariepy

this is the part that im working with, i need to position the piece so that the bottom tab of the piece is always facing a certain position.

David_Gariepy

@Sergio_Rendon,
When you have a part like this where only a small feature give you it's orientation, try to teach only the feature itself. 
In this case, i would try to teach only the tab. If you get false detection, you might want to use the polygon tool to select more then 1 tab.

 

Sergio_Rendon

@David_Gariepy
hi i´ve tried to teach the part using automatic teaching and it recognizes all the tabs on the part, but when i press calibrate, the robot moves to various positions where it cant recognize the part

David_Gariepy

Hi Sergio, 
In this case, you could try to paint the top of the part in white for the teaching process. It will contrast with the inside of the part and you should get a better result. 
Do you have a CAD file or drawing of this part you could send me? I could 3D print it and test it.