Retrofitting machine vision

It can be very tempting to strap a smart camera onto an existing piece of manufacturing equipment, but is this a good idea? From a lean manufacturing perspective, it makes sense to perform the inspection immediately after processing. That way, if there’s a process problem we’ve only made one bad part and not a whole bin or conveyor full. From a machine vision perspective it makes life easier too. Rather than trying to inspect a long list of features or characteristics, we need only look at a few.

Unfortunately though, there is a downside: part presentation. If the manufacturing equipment wasn’t designed to put a part in front of camera there’s a very good chance that it won’t do it very well. For example, if you decided to mount a camera over a conveyor you might unearth several problems:

• The conveyor color or reflectivity might vary, or the surface might be scratched up.
• It might not move at a constant speed.
• The parts might not be placed very repeatably and could wander over the surface, requiring a larger field of view.
• The belt might have some vertical movement, taking parts in and out of focus.

Prevention is usually better than cure, so if you’re considering a vision addition, I suggest this: qualify the performance of the motion or part handling system. Take measurements to understand how consistently you’ll be able to place the part in the field of view. Then, armed with data, decide if it’s good enough or if you need to carry out more significant reengineering.

Chances are, what’s already in place isn’t good enough to support a vision system. Wouldn’t it be good to know that before you spend any money?