This is a major step toward fully autonomous material flows, as envisioned in dark factory scenarios.

After all, robotics is at the heart of dark factories, i.e., factories that – at least in theory – operate without humans and therefore also without light. Repeatable, quality-critical work steps are standardized by robots, sensors, and software in such a way that systems produce autonomously 24/7, as in Xiaomi's smart factory in Changping. There, the HyperIMP AI platform monitors every aspect of production and coordinates eleven robot-controlled production lines that produce a smartphone every three seconds without any human involvement. In practice, we now mainly find so-called “lights-sparse” setups, in which humans only intervene when maintenance, setup processes, audits, or exceptions require it.

Rays of hope thanks to the dark side of manufacturing

The main drivers of the trend toward dark factories are the shortage of skilled workers, volatile demand, rising quality requirements, and the increasing importance of energy and space efficiency. Industrial robots combined with machine vision, IIoT networks, AI controls, manufacturing execution systems (MES), and integrated digital twins form the backbone of dark factories.

Robots take care of their own offspring

A very prominent, often-cited example is FANUC in Japan, where robots take care of their own offspring—in other words, robots build robots. The operation can run for long periods without personnel – the operators speak of up to 30 days at a time. This is achieved through highly standardized processes, predictive maintenance, and controlled environments. In this environment, humans are usually only responsible for regular inspection of the equipment.

Staff are being repositioned

Many years ago, Philips set up a largely dark production line for electric shavers in Drachten, the Netherlands: more than 120 industrial robots, including SCARA and 6-axis robots, carry out assembly and inspection, while a small QA team performs final checks. So “dark” does not necessarily mean “without people,” but often means that personnel are repositioned—in quality, process design, and plant maintenance.

Shipbuilding as one of the most demanding industrial environments in the world

A pioneer in the field of cognitive robotics from Germany is the Metzingen-based company NEURA Robotics. This summer, the Swabian company entered into a strategic partnership with HD Hyundai Samho and HD Hyundai Robotics to jointly develop and test special four-legged and humanoid welding robots for use in the shipbuilding industry. The cooperation is a significant step toward further automation and new robotics innovations in shipbuilding. “With this strategic partnership, we can once again demonstrate the versatility of cognitive robots – in one of the most demanding industrial environments in the world,” explains David Reger, founder and CEO of NEURA Robotics. “Together with Hyundai, we are laying the foundation for completely new approaches to intelligent automation.”

Digital Lighthouse is smart instead of dark

In Germany's industrial reality, smart is currently the norm instead of dark. At Siemens in Erlangen, for example, digital twins link products, lines, and logistics. Simulations and AI applications in Bavaria are improving cycle times, line balancing, and energy use so efficiently that the factory in Erlangen has been awarded the title of “Digital Lighthouse.” This is because it reduces start-ups, increases variant flexibility, and raises overall equipment effectiveness (OEE) without removing humans from numerous links in the value chain.

Keeping production conditions stable while reducing energy consumption

Things aren't all doom and gloom at Tesla either: in its 2023/24 Impact Report, the company reports that most of the HVAC infrastructure at its Gigafactory Nevada is now controlled by AI – with the aim of keeping production conditions stable while reducing energy consumption. In doing so, the US company is targeting a key aspect, as energy is one of the biggest cost and CO2 levers in automotive manufacturing.

Highly varied small series remain challenging

Highly varied, small series – keyword: “high-mix/low-volume” – remain particularly challenging, because according to most experts, so-called dark operation is only worthwhile here with modular plant and software design. Best-in-class examples typically rely on step-by-step automation, digital twins, AI-supported process control, and upskilling of employees. This means that fully dark factories currently only make sense in the manufacture of highly standardized products. In the future, the dark factory will probably serve primarily as a dark background that shows off the shining examples of successful human-machine interactions to particularly good effect.