Value Creation in the Self-Driving Company, Part 2

The first part of this article covered the two drivers at the start of the value chain: research and development, and production forecasting and planning. This second part deals with the three drivers where value creation becomes physical: the automation of production, robotics with digital twins, and automated warehousing. This is where the largest investments are made, and where the difference between the self-driving company and the Industry 4.0 factory shows most clearly.

Five drivers, one chain

As a reminder: a self-driving company uses intelligent software to plan, document, and monitor every step from product idea to finished product, whatever the product category. In my book "Das selbstfahrende Unternehmen" (Springer Gabler 2021, chapter 6.3), I derived five drivers of value creation from this:

1. Research and development
2. Production forecasting and planning
3. Automation of production
4. Robotics and digital twins
5. Automated warehousing

The first two drivers work with information: they decide what is produced and in what quantity. Drivers 3 to 5 turn those decisions into material, machines, and movement. The distinction matters because information can be copied almost for free, while machines and factory halls cannot. Whoever invests in the physical drivers commits for years. All the more reason to know the direction.

Driver 3: Automation of production

Not every core process is fundamentally changed by digitalization and artificial intelligence. Steel will continue to be produced according to a proven pattern, and a cow will still grow best on fresh grass. Technology supports and optimizes such forms of production; it does not replace them. The prerequisite is a certain size of production unit, because automation only pays off above a certain level of utilization. That size does not have to sit within a single business: for decades, the machinery cooperatives known in Austria and Germany as the Maschinenring have shown how farmers share expensive equipment. A self-driving agricultural machine costing several million euros is beyond the reach of a single farm; in a collective, the productivity remains attainable.

Services are a different matter. Some will disappear from the market, new ones will emerge. The book uses passenger transport as an example: the future of taxis and trucks lies in self-driving systems. Two years on, a sober look at the state of play is in order: Waymo carries passengers in Phoenix and San Francisco with no one behind the wheel, while Cruise had its permits for driverless operation in California suspended in October 2023 following an accident. The technology is further along than many expected; the road into regular operation is longer. That changes nothing about the direction. The self-driving vehicle becomes an asset that works: it takes its owner to the office in the morning, transports other people during the day, and, when plugged into the grid, selects the cheapest tariff itself or feeds in the solar power from its owner's roof.

In this world, humans become providers of high-quality, empathetic services. Those who can afford it will still have their hair cut by their hairdresser, even though robots exist for the job, and will pay a multiple of the automated service for it. We will be able to afford that because value creation has risen so sharply elsewhere.

I consider two shifts to be the most important:

• From service to virtual product. We used to ask an expert; today we look it up on Wikipedia or watch a video when the stroller will not fold in a way that fits into the trunk. Irregular tasks in companies will be assisted following the same pattern: maintenance technicians see on site, through smart glasses, what needs to be done instead of being trained for every machine type. The instructions come from the digital twin, more on that shortly.
• From physical product to software product. An electric motor does without hundreds of moving, wearing drivetrain parts. The physical car becomes cheaper as a result, while more and more software is built in; assistance systems such as Volkswagen's IQ.DRIVE show where things are heading: voice control instead of knobs, until at some point the driver is no longer needed. Value creation in the automobile is migrating into software, not least because fully automated manufacturing makes the vehicle itself ever cheaper.

For companies that manufacture physical products, this means that the share of value creation that comes from software rises with every product generation. Anyone who regards software merely as an IT cost center is missing that part of the business.

Driver 4: Robotics and digital twins

Today's factories operate a multitude of individually programmed robots, above all stationary robot arms. They are precise but immobile: it is still the people who move through the hall as needed, increasingly supported by automated logistics systems. Autonomous mobile robots such as the Agilox vehicles developed in Upper Austria already supply production lines with parts on their own, avoiding people and obstacles as they go.

The factory of the future goes one step further. As Tesla's Gigafactories demonstrate, it becomes programmable as a whole: algorithms continuously reconfigure it according to current requirements, and standardized, highly flexible robots are deployed wherever they happen to be needed. Even the instructions for how a work step is to be carried out are generated automatically. The tool for this is the digital twin.

The concept goes back to Michael Grieves, who introduced it in 2002: a virtual copy of a physical object that remains connected to the original in real time via sensors. The machine's data is evaluated and simulated in the twin; a maintenance procedure is captured and digitized immediately; the data resides locally, in a decentralized setup, or in the cloud. Since 2021, the ISO 23247 series of standards has described a framework for digital twins in manufacturing, a sign that the concept has left the laboratory. For manufacturing, maintenance, and servicing processes, digital twins become indispensable because they shift trial and error into simulation: a changeover of the line is calculated on the twin before a single robot is moved.

Communication between robots and people is based on algorithms and deep learning and will be as natural in 2035 as the smartphone is today, which likewise went from novelty to everyday object in roughly ten years. Many households already show what this looks like: voice assistants control heating, shading, ventilation, or vacuum cleaners, recognize the voice for authorization, and link the subsystems into a whole to which the residents communicate their wishes. What works in the living room also works on the factory floor.

Driver 5: Automated warehousing

An automated warehouse is essentially one large robot with clearly defined tasks: storage, retrieval, and relocation run autonomously, and picking follows the goods-to-person principle. Conveyor technology brings the items directly to the picking station instead of people walking the aisles with a cart. The person at that station will be a robot in the future. The advantages over the conventional warehouse are tangible: less floor space, less energy, shorter distances, and shorter access times, because algorithms continuously calculate the most economical overall solution, plus integrated material flow control. That this expertise is at home in Austria is shown by providers such as TGW and Knapp, which deliver warehouse automation worldwide.

During the transition period up to 2035, warehouse automation has a further effect: it relieves employees of physically demanding and monotonous work steps, and the errors that inevitably occur in manual picking fall to a minimum.

Technically, this rests on robust structures, energy-efficient conveyor technology, and cloud-based warehouse management software that controls all operations. The warehouse management system controls inventory and material flow and is therefore rightly called the pacemaker of the warehouse. In the self-driving company, the warehouse is moreover connected to all other business functions, with continuous mutual coordination in real time:

• Orders are triggered automatically as soon as inventory and forecast require it.
• Invoices are issued and checked virtually.
• Continuous inventory counting gives controlling up-to-date data at any time, from which decisions are derived, such as negotiating new supply contracts with new partners.
• Internal data is linked with external data and forecasts: the company knows exactly what was sold in the last period, from the year to the quarter to the last second, and what demand to expect in the next.

Management no longer intervenes in individual orders. It defines system boundaries and receives notifications or warnings when the otherwise fully self-driving warehouse operates outside those boundaries. This is the meta-decision I described in the article How Self-Driving Companies Decide, applied to the warehouse.

The automated warehouse shows how the rigid boundaries within the company and toward its environment break open in favor of an overall organism that adapts in real time.

This interaction improves the more partners are self-driving as well: the logistics company that provides delivery information in real time, or the supplier whose production accepts the order without a detour through a mailbox. The self-driving company is not a closed system but a node in a network.

What the three drivers have in common

With none of the three drivers does value creation come from the individual machine. Robot arms, automated guided vehicles, and high-bay warehouses have existed for decades. What is new is the software that connects them with each other, with the forecasts from driver 2, and with partners outside the company, and that continuously calculates the best overall solution from the data. Manufacturing costs then consist essentially of one-time investments plus ongoing energy and material costs; the share of manual work falls, and the remaining human work migrates into development, monitoring, and the exceptions that no algorithm anticipated.

From our projects, I derive three recommendations for getting started:

1. Start with the warehouse. It has clear boundaries, measurable key figures, and, in the warehouse management system, a piece of software that is needed anyway. Hardly any other area can be delimited so cleanly and calculated so quickly.
2. Build the twin before the robot. Whoever captures and simulates the data of existing equipment knows which automation pays off before the first machine is ordered.
3. Plan the connections, not the islands. Every automation island that does not talk to planning, purchasing, and partners merely shifts manual work to its edges. I described the architecture for this in the article Enterprise Architecture for Self-Driving Companies.

The next step

The five drivers of value creation are described in the book "Das selbstfahrende Unternehmen" (Springer Gabler 2021) in the context of the model's other building blocks: About the book. If you would like to know which of the three physical drivers offers the greatest leverage in your company, we would be glad to discuss it: Book an expert consultation.