Automating Visual Inspection in Copenhagen

This week the big thing taking up a lot of my time was the paper I had to write for the 2012 Turbo Expo conference in Copenhagen.  The topic was automating Visual Inspection.  


The most memorable quote I came across was from a study “ Visual Inspection Research Project Report on Benchmark Inspections” by Floyd W. Spencer.  At the beginning of the paper he adds to the definition of visual inspection saying not only involves the use of the eye but shaking, listening, feeling and sometimes even smelling the aircraft and its components.


Wow, how do you expect to automate that the skeptics are screaming!  And yes I started to sweat a bit thinking I could get that question as I present the paper in front of a bunch of design engineers!  


Automating detail processes that require an intuitive approach is not easy.  But as anyone used to automation does you pick your battles, make your assumptions and layout your plan-of-attack focusing on what can you take on.


The rewards are worth it.  80% of the inspection done on aircraft components is still visual inspection.  And the toughest part of bench marking human visual inspection is the fact that you can't.  Everything depends on the unique site where the person is inspecting the parts.  It depends on the training, environment, the parts, their mood, time of day etc etc.


So definitely the rewards of developing a technology which is linked to quantifiable standards will be a great relief to any company who has been on 3rd party containment or any manufacturing engineer who has had to chase down serial numbers of escaped parts before they get into an engine.....and if they get into the engine the rewards grow exponentially!

3D Image from Surface Profiler

Of the projects I've worked on the toughest part isn't implementing the calibration routines or algorithms to ensure all defects are detected, it is educating the end-user on the system they are receiving and  how to effectively use it in their manufacturing.

All of a sudden they are bombarded with new and important information on their parts.  Normally this was lost because of documentation practices or the fact the final inspector wasn't an inspector, but a rework artist.  So there is a step change required and a team effort that involves people outside of the inspection area.  This data can be used as feedback to improve the manufacturing process to ensure the parts getting to final are actually final parts.

If you target in-line inspection you can reduce the number of rework loops that happen because of the defects that are missed on the first inspection.  But what about the defects that are created by the rework operator?  Is this the next area that should be automated?

www.avr-vr.com
http://ca.linkedin.com/in/muldoonmichael