Machine development:

Will the driver remain the determining factor?

Agricultural technology has followed the paradigm of "everything for the driver" for 100 years. What started with a seat, steering wheel, pedals, a few levers and switches has now expanded to whole machine designs and processes. However, "everything for the driver" now entails a path dependency that we should question for once: immense development effort and production costs for the machines, even greater material consumption. Soil compaction on the field and, if everyone drives as they please, danger on the village road as well. How much is invested in comfortable cabs, in sound insulation, air conditioning, suspension and safety, in internally ventilated super-chairs, armrests and joysticks, screens and software! This draws the full attention of the developers, but creates new problems with every solution, which are then often even outsourced (passed on?).

Because one person is expected to do more and more, working width and power have become the measure of all things. But do you really need 700 hp and 14 m to cut a few stalks and thresh out their ears?

Everything for the driver has been the credo in agricultural engineering for decades. But what can be done when there are fewer and fewer drivers? Photo: landpixel
Everything for the driver has been the credo in agricultural engineering for decades. But what can be done when there are fewer and fewer drivers? Photo: landpixel

The problem of soil compaction

After years of appeals, the problem is finally being seriously addressed with track technologies and air pressure regulations. But isn't that an expensive rework? The weight has to remain, because otherwise the 6 or 12 m wide cultivator cannot be pulled properly. 12 x 1 m is not possible at all? Here, too, path dependency is a strong driver for optimising functions: what has always been done now works a little better, faster, more compact, more powerful, more efficient. At Agritechnica it says "NEW", but it's not different, it's just more of the same.

Automation serves the driver first and foremost

Automation simplifies processes because fewer and fewer well-trained people want to "sit on the goat" and spend twelve hours pulling lines in the infinity of the Kazakh grain region. Automation helps because machines are becoming more and more complex and even professionals who reactivate the thresher after eleven months in the barn ask themselves: What was that about the separation optimisation menu?

Many automations also bring real productivity gains. Steering systems can save 10 to 15 %, because even experienced drivers cannot avoid overlapping with ever larger working widths. TONI and TIM finally made the tractor not only "listen" to the implement without detouring via the driver, but also implement the driver's will. Headland management prevents breakage and soil damage and optimises turning time.

Tired or inexperienced drivers are often the weakest link in the process chain. But what to do if drivers are missing in the first place? Agriculture is also practiced in culturally and infrastructurally lonely areas. In the truest sense of the word, humanity lives from the fact that it is precisely there that farming is done effectively.

Things are always moving forward. But perhaps also into a cul-de-sac? Photo: landpixel
Things are always moving forward. But perhaps also into a cul-de-sac? Photo: landpixel

An interim conclusion

Functional optimisation has phases that bring added value for all without major side effects. But it reaches its limits when each advance brings ever smaller added benefits with ever greater (also financial) efforts at the same time. More and more participants in the development processes and later even the users have doubts. This is where we are right now, because something significant is happening.

Automation is followed by autonomisation as a logical further development. New sensor technology and new imaging processes are increasingly better able to be used "remotely" on the machine or via high-performance radio networks with their flood of data. Clever software that uses the first two stages of artificial intelligence - pattern recognition and machine learning as well as the first stimulus-response processes - make complex technical systems more suitable for everyday use and more widely applicable. Even amateurs can operate them. In the process, "operation" no longer tells the truth at all, because the machine conveys to the driver: "Drive me to the edge of the field, but then keep your hands off me!" So when the assistance role of partial automation moves on step by step to full automation, the old paradigm of "everything for the driver" starts to hiccup. Because every expensive further development has to ask itself the question: "What does the driver do then?" As long as he cannot (may not?) be sorted out of the process costs, the full autonomy business models do not pay off. There still has to be a driver infrastructure.

If automation means that more and more parts of the process run without a driver, then autonomisation consistently means "EVERYTHING without a driver". No remote control, no monitoring at the edge of the field, no human being in the process. At least none that would be necessary.

Excursions into the unknown

The "makers" of innovations show you perfectly what they have changed or want to change. Starting from the functionality of their machine, they assess the surrounding environment - depending on their openness and competence, only the obvious or sometimes a bit more complex. The briefing of their management has probably been for years: "How do we get growth?" A question they obviously ask themselves too little, on the other hand, is: "Where does growth come from?" For this they would have to organise excursions into unfamiliar subject areas, orient themselves under uncertainty and sort out observed approaches at great expense. Controllers and many managers, especially in efficiency-driven companies, hate this approach because it can be so ineffective, so time-consuming. It swallows up time without any demonstrable benefit.

But engineers would also have to deal with topics that require them to quickly acquire knowledge they do not have yet. They would have to be able to classify whether what they are looking at makes sense, is well thought-out, forward-looking and attractive for more in-depth examinations. Then they meet AI experts fresh out of university who (still) lack sound agricultural expertise, and the reservations are already greater than the curiosity. Or they meet lateral entrants from other sectors - and far too quickly the barrier is in their minds that in agriculture, contrary to all other sectors, the unpredictable nature is the main complexity booster and therefore everything is different.

Yet it is true that only those who know what they might want can even think about what they want.

Henning Rabe, 

PHAEN Agtech, Kiel