Regardless of the level of sophistication of your systems or the complexity of operations, there is no easy way to monitor a work center today. Enterprise Resource Planning (ERP) systems push work onto the floor in batches or via kan bans. Manufacturing Execution or Shop Floor Control Systems help control and schedule work around available people and machine capacity. The problem is that none of these systems actually watch the action on the floor. That is why the first lesson we were taught in operations is that there is no substitute for managing by wandering around (or MBWA).
When I was just starting to run shop floors as a materials and supply chain manager, a very wise plant manager shared with me his daily route to stay on top of things. While his subordinates all parked in the front of the plant near their offices, he parked out back by the shipping and receiving docks. He made a point of taking a serpentine route from the back to the front of plant. In this way he could detect any bottlenecks or other issues in shipping, receiving, stores, and every work center. By the time he got to office, he had a pretty good idea of what to focus on for the day.
The problem with this technique, or any MBWA for that matter, is that it can only provide a snapshot in time for each area observed. If a problem occurred after he walked by, he would be typically clueless until he read a production report at the end of the shift or the next morning. I used to love wandering around as a way to meet the folks on the floor and learn how things were made. While it is a great learning tool, MBWA cannot effectively manage a busy operations floor.
Computer vision offers a means to radically update and improve management by wandering around. Unlike the plant manager’s snapshot view of how people, equipment, and products were interacting, computer vision based on artificial intelligence watches every person, every piece of equipment, and the products and materials they are working on. All this is continuous and in real time.
Unlike a surveillance camera system that requires a human to view hours of video feeds, computer vision automates continuous monitoring and only sends alerts when a problem occurs such as an ignored machine, or a production line that stops moving. All the action is captured for further analysis and continuous improvement efforts.
Computer vision also overcomes the age-old Hawthorne effect in which people act differently when being watched and measured. Typically, the operation improves as the surveillance continues, but then reverts back to earlier, less productive levels after the observers depart. This a big reason why a majority of Lean and Six Sigma projects are deemed a failure by their sponsors. I should know, I ran over 20 such projects over a six-year period.
Computer vision can even automate paper-based work orders systems. Using optical character recognition, work orders can be read when placed in a designated basket or tray. This is especially helpful for smaller operations that want to avoid the cost of implementing (and maintaining) expensive manufacturing execution systems. It avoids problems with data entry errors and reduces information latency as well.
Now for the best part. With continuous monitoring and alerts based on computer vision, you do not need to be on site to manage your floor operations down to individual work centers. Not only will you have control of daily operations, you will have the ability to analyze them, and make continuous improvements truly continuous.
I only wish my old plant manager mentor could be around today to run his operations sitting in an easy chair on the beach.
