Utopian visions of a 'dark factory', in which machines and artificial intelligence control factories and production processes without human intervention, are driving the imagination and fears of many industrial decision-makers and employees. In this context, the 'dark factory' is often presented as a vision of the future that manufacturers should strive for. But is this concept, which essentially involves production without people, a realistic and desirable goal at all?

Fully automated factories such as the Foxconn plant in Shenzen, China, which has been recognized by the World Economic Forum as a leader in the fourth industrial revolution, already exist. However, such examples are rare. Dark factories are most popular in areas where the aim is to minimize human interaction, such as production processes with a significant risk of contamination, for example in food processing or the manufacture of special electronic components. Also of interest are areas where product margins are higher, such as in pharmaceutical production, where a return on investment (ROI) is easier to achieve.

However, these applications almost all require a high-volume, low-product-mix scenario in order to be profitable. This is at odds with the general move towards low-volume, high-product-mix in many value chains, as companies want to create more choice for their customers. Therefore, the 100% automated, autonomous, completely unlit and unmanned factory is not a realistic goal for most manufacturing companies.

Rather, there is an optimum for every manufacturer between 0 and 100% automation at which they can achieve the best return on investment in automation technology. If this point is exceeded, profitability decreases as automation progresses. A Gartner study from June 2020 came to the conclusion that only 17% of respondents believe that they will fully rely on digital, comprehensively automated production facilities by 2025. The majority (79%) assume that their production processes will be based on human-controlled processes that are meaningfully supported by digital processes.

One of the main reasons why a fully automated factory cannot be realized is that production requires human skills. As advanced as today's automation technologies may be, there are situations where human flexibility is still best suited to carry out complex processes: For example, when improvisation or complicated decision making is required to respond to an unexpected event. A good example is processes in a bottling plant: if a bottle jams, it is not easy to fix this automatically. For a human, on the other hand, it is a fairly simple problem that is easy to solve.

Without human employees? Not possible!

Digitalization makes it possible to capture valuable process information that can be analysed to identify the root causes of production problems. This can then be used to adapt lines and machines in order to minimize or even avoid machine downtime. To stay with the example of the bottling line: Here, it would be possible to redesign the filling machine's infeed flows to eliminate the bottle jam. However, if a subsequent batch of bottles from the supplier does not meet the specifications, the problem could reoccur despite all efforts. Although steps could be taken to measure the bottles before they are fed into the line, a material problem could then occur, causing another machine error. So basically, steps can always be added to optimize a line. But no one can really guarantee one hundred percent that all potential problems will be eliminated.

Examples like this show that it may not be practical to automate the last 10% of a production line in order to realize the complete 'dark factory'. Irrespective of this, however, a general trend is for employees in production to increasingly take on supervisory functions instead of traditional operational tasks.

'Proof of concept' brings clarity

A crucial step for manufacturers is therefore to find out to what extent and in which areas automation offers the optimized return on investment. Many system integrators and technology providers help to carry out an analysis and realize a 'proof of concept' - from individual applications to entire systems. Omron, for example, has dedicated proof-of-concept labs in Stuttgart, Dortmund and Langenfeld as part of a global infrastructure of technology centers. These locations are regularly used to construct prototypes that enable physical demonstrations as part of the proof-of-concept process.

From image processing to robotics

Advances in automation technology mean that production processes are steadily approaching full automation. One example is industrial image processing, where increasingly powerful 3D vision systems are enabling ever more complex pick-and-place processes. Whereas components used to have to be transferred precisely, modern image processing technology can easily identify objects that are randomly arranged in transport containers.

A fundamental requirement for the automated factory - and crucial for 'dark factory' projects - is robot technology, specifically the transportation of materials, sub-assemblies and other objects around and within the production process.

Whereas in the past, various types of guided industrial trucks and forklift trucks were used, today mobile robots can automate such tasks. They are controlled by a higher-level control system that ensures that the right materials are delivered to the machines, work in progress is transferred between production systems and finished goods are returned to the warehouse. Mobile robots can carry considerable loads and work seamlessly with each other and safely with humans.

Cobots in high-mix, low-volume use