Can the camera be used to identify parts in a box?

Tim_Smith · November 2016

Hi DoF community,

I was wondering if the Robotiq camera can be used to identify parts inside a cardboard box? I working on an application where the parts are in trays that are stacked inside a cardboard box. The process consists of picking the parts out of the box and placing them in a CNC machine. So the robot would have to pick out the parts, then remove the top tray to access the lower trays. Is lighting going to be a problem here?

Etienne_Samson · November 2016

@Tim_Smith shouldn't be a problem but remember that the camera is 2D, thus you will need to build your program with a multi-level structure for snapshot position, going down through your trays one by one. Each layer should have it's own snapshot position with calibration.

There is a discussion similar to this, to empty pile of trays : http://dof.robotiq.com/discussion/357/emptying-pile-of-trays-with-vision-system

Another thing, we do not have collision detection, so if you have a tight environment (the box) watch out for collision with the sides or with other parts.

Tim_Smith · November 2016

@Etienne_Samson Thank you for the insight! Do you think the camera would be able to identify the parts at different heights if the height only varies by a few inches? This way I could use one camera locate and change the MoveL inside the camera locate node.

Etienne_Samson · November 2016

@Tim_Smith It will be able to identify them up to a certain limit, but the more difference in height you have the greater will be the imprecision on the position. If you got an application that doesn't require precision at all you can probably go with a single snapshot position, but if you need any kind of precision your parts will have to be on a plane linked to the snapshot position.

Discussion

Can the camera be used to identify parts in a box?

Comments