A very hot concept in manufacturing is the prospect of “Industry 4.0”. Originating in Germany, it is envisaged as a factory where recently emerged technologies (automation, digital communication, Internet of Things, cloud, etc) are combined to assemble complete, automated production systems that require minimal human oversight and intervention. The following YouTube video makes the concept seem pretty real.
What does Industry 4.0 involve?
Pretty much every major industrial vendor and consultant has published serious views about Industry 4.0 in the past few years. These are a sampling of their PDF summaries here:
Industry 4.0 focuses on the physical and IT building blocks of an end-to-end manufacturing system that operates with far less human intervention. Definitions vary, but most revolve around lists or descriptions of the technologies that will be used and supported. One of the most expansive lists is offered by McKinsey.
Where does video fit in?
Apparently, it doesn’t.
Mention of video technology is strikingly absent from the various industry definitions. As an exercise, try to find it in McKinsey’s chart. Video is a dynamically changing technology that is reshaping huge parts of the social landscape. Yet, it is not evident in the discussion of Industry 4.0. Take recent PDF whitepapers from the major consultants and do a text search for the term “video”. … pretty thin pickings!
I plan to write a series of blog articles about the details of a possible role for video technology in Industry 4.0, but to keep this article short and to start the discussion, I want to frame a big-picture rationale for why we will still need to see our shiny, digital factory … with our eyes, not just our sensors.
In many previous posts, I have argued that we should use video far more frequently to monitor and study business operations. Nowhere does it make more sense to me than in factories. However, I suspect that part of the reason we don’t see heavy video use now is because our factories are heavily staffed with human eyes and ears. Every manufacturing expert since Deming has tasked front line workers to be alert and recognize when odd events occur, so they can respond quickly and/or tell management. Good factories rely on the eyes and ears of its people to know what is happening. Video surveillance would certainly be useful, but companies can get by as long as the plant is fully and actively staffed.
However, Industry 4.0 expects to see a big reduction in plant floor headcount. There will be fewer eyes and ears. Look at this photo that comes from the video at the top of this post. It was taken in the Siemens plant in Amberg, Germany.
- How many people do you see?
- How many things must be happening simultaneously that this staff cannot possibly watch.
- If something unexpected did occur, what are the chances they might have seen it?
The concepts for Industry 4.0 assume that sophisticated software controls will apply to every aspect of the process. There will be tons of sensors, software and sophisticated diagnostics. Presumably, the factory management software will recognize problems and respond faster than humans could. All of this assumes that the designers can anticipate the potential failures. If not, the limited staff will become detectives. They will have to piece together the root cause of each event … without the help of witnesses.
I expect it works reasonably well for products that are hard-bodied, modular, discrete and easily tracked. In other words, I take Siemens at its word that it works for the electronic products that are made in the Siemens’ plant. That will also apply to a lot of products in automotive and aerospace industries. What about softer products like upholstery in an auto plant? What if there were a lot more bits and pieces that were harder to track … like fasteners? What about wires and stuff? What if the raw materials are harder to standardize? What if … ?
Maybe it’s just me, but I would love to have a robust video surveillance system watching this factory, both from a long view and with less expensive cameras in closeups of key machines and production steps. If we specified 200 cameras, including some 4K high def units, I think we could we could aim them at every point where a failure or issue might occur in this factory. We could record everything that happened, 24/7, with a rolling 2 or 3 week buffer.
How much would that cost? Hardware from a high-end vendor like Bosch would probably come in around $120 to $140K. Installation might be another $50K. Do you think that would cause heartburn for someone building a plant like this?