Industry 4.0 Is Over, Now Comes the Self-Driving Factory

For a good ten years, "Industry 4.0" has stood for the networked factory: sensors, real-time data, and robots that tie production and the supply chain more closely together. That was a necessary step, but not the destination. The self-driving factory goes further: it is programmed rather than built, it configures itself according to current demand, and it optimizes itself as a whole rather than department by department. This article puts into perspective what Industry 4.0 has achieved, what comes next, and how you can tell whether your plant is ready for it.

Every industrial revolution has redistributed value creation

Looking back at the industrial revolutions, a pattern emerges: each of them was driven by an increase in value creation, and each redefined the role of people in the factory.

• The steam engine. The first leap concerned logistics. Machines took over the transport of raw materials, steel parts, and finished goods, which until then had been handled by people, horse-drawn wagons, and ox carts. Transport became faster, safer, and soon cheaper.
• Mass production. With the second revolution, machines entered manufacturing itself. In 1913, Henry Ford introduced the assembly line on a large scale at Highland Park, and Taylorism supplied the calculation methods to go with it. Factories were completely restructured; people worked at fixed stations, performed the same movements ever faster, and routine brought monotony.
• Digitalization. The third revolution transferred mental work to machines for the first time. Many new activities emerged, but so did new forms of monotony: the same spreadsheet analysis over and over, which is still everyday practice in many companies.
• Industry 4.0. The term was presented in April 2011 by the Communication Promoters Group of the German Industry-Science Research Alliance. The fourth revolution automates coordinating and communicating processes, previously handled slowly and error-prone by clerks and middle management, in real time and beyond company boundaries.

Just as constant as progress were the fears that preceded it. In the 19th century, rail travel above 50 km/h was considered a health hazard; when machines entered the factories, people feared for their jobs; the same repeated itself with the first PCs in the office and with the internet around the turn of the millennium. None of these fears came true. What did disappear were the companies that clung to old methods: the wheelwrights and blacksmiths, the mills, the typewriter manufacturers, and most recently Nokia and Kodak.

What Industry 4.0 has achieved and where it ends

Industry 4.0 equipped the factory with sensors, intelligent algorithms, and the analysis of large volumes of data. When a vehicle is sold in Vienna, the information goes automatically to suppliers worldwide, who adjust their lines and capacities. The vision of lot size one, the custom product at the cost of series production, has arrived in software-supported manufacturing. Evaluations show that introducing these technologies can reduce costs in manufacturing companies by 15 to 20 percent (Obermaier 2019, cited in "Das selbstfahrende Unternehmen", Springer Gabler 2021).

It has come at a price: over the past decades, mid-sized companies in Germany, Austria, and Switzerland have invested very heavily in production and value-creation processes and neglected the supporting business areas. To this day, sales and production plan largely separately; the information gathered by the field sales force never reaches manufacturing.

More importantly, Industry 4.0 optimizes within a structure designed by people. The layout, the sequence of work steps, the assignment of robots to stations: all of this is planned, documented, and re-planned by experts when necessary. The programmable robot arm is a good thing, but it stays fixed in place; it is still the people who have to move. The technologies for a fully automated, "transparent" factory have been available for 15 to 20 years, as set out in "Das selbstfahrende Unternehmen". Few have implemented them so far, because the combination of human labor and partial automation was cheaper than a complete fleet of robots. I expect this ratio to tip in the coming years.

The self-driving factory: program it instead of building it

The self-driving factory is the logical continuation of Industry 4.0 and digitalization, and it still follows the classic principles of industrialization: automation, standardization, modularization, specialization, continuous improvement. What is new is who runs the factory. Formerly isolated production units and linearly organized processes become holistically networked, multidimensional functions that optimize each other in real time with respect to their system status by means of self-learning algorithms. Bottlenecks, idle times, and maintenance intervals are reduced to a minimum; manufacturing costs consist essentially of the one-time investment and the ongoing cost of energy and materials, while personnel costs fall drastically.

Tesla shows what this looks like in practice. While the established manufacturers adapt their production step by step on the basis of existing technologies, Tesla took a different path: not building a factory, but programming one. The basic idea is simple: what has been programmed once can easily be changed. Just as the vehicle's chassis is retuned via software update, the Gigafactory adapts to current demand. The robot systems in use accommodate qualitative as well as quantitative changes, adjust to rising sales, and the factory thus scales like a pure software startup. And because the principle is programmed, it can be copied: the Gigafactory built in Nevada from 2014 has been followed by plants in Shanghai, in Grünheide near Berlin, and in Texas; the latter two opened this year.

A factory that is programmed rather than built can be changed like software and copied like a blueprint.

The approach that was initially ridiculed is now putting the major carmakers under considerable pressure. This is not a question of industry but of architecture: any industrial company can design its production as a programmable system.

Three building blocks of the self-driving factory

Flexible robotics instead of stationary arms

In most factories today, a large number of individual programmable robots are at work, each at its own station. In the self-driving factory, all physical work is carried out by standardized, highly flexible, and interconnected robot systems. Algorithms continuously reconfigure and reprogram the factory based on current requirements; the robots are deployed wherever they are needed at the moment. The factory as a whole thus becomes a programmable object.

Digital twins close the gap between real and virtual

How the individual work steps look is likewise prepared automatically, for instance by means of digital twins. The digital twin is the virtual replica of a physical object; the term was coined by Michael Grieves in 2002. Sensors on the real object deliver data in real time to the virtual copy, where it is analyzed and simulated. A maintenance procedure on a machine is thus recorded and immediately available digitally; manufacturing, maintenance, and repair processes can be improved on the twin before anyone touches the real equipment. Where observation and a stopwatch were once the basis of every improvement, everything is now available as data. With growing computing power, more and more functions of the programmable factory will build on such replicas.

The warehouse that manages itself

Automated warehouse systems work in principle like robots: storage, retrieval, and relocation happen autonomously, and the goods come to the picking station, not the other way around. Compared with the traditional warehouse, they need less floor space, save energy, shorten distances because the algorithms continuously calculate the most favorable overall solution, and offer shorter access times with integrated material flow control. The warehouse management software is the control center: it controls stock and material flow, triggers orders automatically, and, through continuous inventory, provides controlling with up-to-date data at all times. Linked with external data and forecasts, the self-driving factory knows what has been sold and what demand to expect in the next period. Management defines the deviation at which notifications or warnings are triggered, and intervenes only then.

The overall optimum instead of the sum of departmental optima

What the factory shows so clearly applies to the entire self-driving company: rigid boundaries are broken down in favor of a hybrid, transparent overall organism that can be adapted in real time and is networked with all relevant actors. The interaction gets better the more partners and suppliers are also organized in a self-driving way.

Experience from analog companies shows why this is necessary. Contrary to the overall optimum, departments tend to pursue only their own optimum. The reasons are deeply human: ambition, self-promotion, the pursuit of power and career, narrow-mindedness. They lead to conflicts, inadequate communication, and the withholding of valuable information. The self-driving factory offers the chance of complete transparency: the data from all areas is continuously evaluated and contributes to optimized functions across the company. In the process, people are relieved of monotonous, tiring tasks: after the assembly-line workers and the clerks, it is now the warehouse employees, and they can devote themselves to empathetic, creative, and steering tasks. Research and development remain a domain of people, supplied with information from production, customer channels, and partner organizations in a quantity and quality that no analog-organized company can match.

What this means for industrial companies today

The self-driving factory is not a project you commission but an architectural decision that pays off over years. From our projects in manufacturing, we see four prerequisites:

1. End-to-end data from sales to production. The sales forecast and production planning belong in one system. Highly digitalized companies already ship precisely calculated quantities to the regions where they are likely to be sold.
2. Algorithms instead of processes. Anyone who continues to model operations as a rigid process chain cannot reconfigure them continuously. I described why the self-driving company builds on algorithms rather than end-to-end processes in the article The End of Processes, Long Live the Algorithms.
3. A software architecture that connects production and business processes. The factory is only as programmable as the systems it is connected to. The foundations for this are set out in the article Enterprise Architecture for Self-Driving Companies.
4. Programmability as an investment criterion. Every piece of equipment, every warehouse system, and every software solution procured today should be assessed on whether it can be reconfigured via software or whether every change requires a physical conversion.

Anyone who takes these four points seriously is on the way. The restructuring this requires is fundamental. But the companies that allow it will be the winners of the next decade, just as the steam locomotive, mass production, and the PC rewarded those who were willing to invest in new technologies.

The next step

The self-driving factory is one chapter of the vision I described in "Das selbstfahrende Unternehmen" (Springer Gabler 2021, in German). You will find an overview of the book in the summary of the book and on the page About the book. If you would like to assess how far your plant is from a programmable factory, book an expert consultation.