You occasionally witness repetitive problems. If you could examine them closely as they happen, you can probably figure out how to deal with them … but they seem to occur randomly and intermittently … typically when no one is watching closely.
A permanent camera is focused on the work area and it continuously records HD video that is stored on a network server. You also have video performance tagging and measurement software (e.g., Dartfish.com).
Search the recorded video for a time segment when you know the problem occurred, ideally where it happened several times in a reasonable time span. Export the video and import it into Dartfish. Using a Dartfish tagging panel, scan quickly (e.g., 4x normal speed) to find examples where it happened and tag the events and their durations. The following example illustrates how this can work.
Export the tagged event sections as a video sequence, where one event plays right after another, then watch the examples, and go back and forth to look for clues and patterns.
Here, if the worker in these pictures didn’t have waffles to place on the bagging line, she reached back to grab them from a nearby buffer. That was her instruction. After reviewing examples of this intermittent event, it was clear that upstream workers usually supplied new stock before she could grab the spares. She would produce more, with less stress, if she just waited for them to catch up. Her instructions were changed.