Poka Yoke for Packing Accurate Box Quantities

Six Sigma – iSixSigma Forums General Forums Tools & Templates Poka Yoke for Packing Accurate Box Quantities

Viewing 4 posts - 1 through 4 (of 4 total)
  • Author
  • #55446


    Hi Everyone,

    I have been working on developing a poka yoke to ensure our boxes are shipped with the correct quantity.

    Our general process is: Assemble part –> Stamp Part –> Place part in bag –> Scan bag label into system –> Place part into box

    Our break down is occurring when one piece flow is not followed. Multiple pieces may be bagged up during the shift and then each bag is scanned in at the end resulting in duplicate scans or missed scans. This translates to our system generating box labels with incorrect quantities.

    I want to implement a sensor or scale that won’t let the operator scan in an additional bag until that sensor is tripped. My issue is that I can’t find a photo-eye sensor that would connect directly to our computer via usb. All it would need to do is send a message (yes or no) that the sensor has been tripped.

    Any suggestions?



    Without knowing your layout, volumes etc this is a tough one to assist with.

    That being said a couple thoughts…
    For the best Poka Yoke/error proofing ideas you should take a step back and ask how you can design the flow so it can’t be done incorrectly. A tripped sensor still may not prevent your errors, just add an extra step in the process since it sounds like one piece flow is not your standard work setup.

    Is there another low cost/no cost way to ensure one piece flow?
    Is there a way that parts and boxes can travel together to ensure the final step, the scan and place into box cannot be incorrect?
    Can you change your process flow to either get one piece flow every time, or if doing batch processing, ensure scanning part to box can’t be messed up?

    Engage your workers and ask them how they might solve the problem. They may already know the answer too and also appreciate being part of the solution.

    Just my random thoughts.


    I would look at developing low cost image recognition. With algorithm trained you can use even RaspberryPi to check what is packed, goes in and interface. The days to have a barcode is not needed – can read text more accurate than having someone scan that – forgetting to scan – or have alignment problems.
    Good luck – see for learning the right stuff.


    Norbert Feher

    Instead of focusing on detection I suggest put Your efforts to prevention.

    There are great ideas for example the above mentioned progressive scaling method, kit to exact or I wonder if You heard about pick to light systems.

    Please key these phrases into Youtube and You will find nice examples

Viewing 4 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic.