Autonomous helpers in the field

We may be familiar with robots from television or the factory, but not necessarily in the field. Between sugar beets, lettuce and co. Whether weeding or sowing seeds, robots are increasingly taking over manually performed tasks in the field. The "Mars rover" in the field promises to make work easier for those involved in agriculture and to conserve resources, e.g. with regard to the use of pesticides. The autonomous helpers can thus make an important contribution to more sustainable and ecological agriculture. How these work and how their use will expand in the future, explain to us Stefan Kopfinger, head of the research group agricultural robotics at the Bavarian State Institute for Agriculture at the Ruhstorf an der Rott site and Christian Metz, head of the Competence Network Digital Agriculture Bavaria - the KNeDL at the ZD.B Center Digitalization.Bavaria - of Bayern Innovativ GmbH.

Robotics in industry is already known to many. But what about out in the field, Christian? Where do robots have a benefit?

Christian Metz: The question is: Are the fields, like the tractors, also getting bigger and bigger, or is the whole thing limited? And so we can simply use the robots to adapt the machines(size) to the landscape again. So the robots can work on heterogeneous fields and also make it easier for us in the way we grow things and also how we work there. The robot replaces or supports my physical, manual labor.
Another point is also that by working specifically in the field, it can distinguish weeds - or as we say, co-weeds - from crops. Thus, he can automatically remove the weeds himself and there can be no need for chemical pesticides.
So it is about the reduction of such inputs that we do not necessarily want to use, both for economic and environmental reasons. And that would be so the big and whole, why the robots will perhaps experience a triumphal procession in agriculture.

What about the acceptance among farmers, Stefan?

Stefan Kopfinger: We have possibly the Bavarian Special Program Agriculture Digital (BaySL Digital). There, among other things, chopping robots are promoted. In the past two years, there have been 64 funding applications for this. In other words, there is interest in the niche area. But at the moment, we are seeing growing interest in systems that replace manual labor.
Consumers don't necessarily care whether a robot is driving around in the field; their focus is on having high quality food. The reduction in herbicide use, for example, through some process facilitation in mechanical weed control is then an advantage for consumers.

Stefan, you are head of the research group Agricultural Robotics. What exactly do you deal with?

Stefan Kopfinger: We are currently mainly concerned with autonomous systems for autonomous weed control, also called hacking robots. The thing about robots is that since there is no driver on it, there is no driver's cab and that's why a field robot can be designed very freely in terms of shape and size. Our smallest robot is powered by 400 watts of electric drive from a battery and the largest by 150 diesel hp. In other words, there is a very wide range in terms of shape. But as a common feature most of them have a track guidance via RTK-GNSS, i.e. a satellite navigation system improved over accuracy.

GPS we know from the car. But this is the much more accurate variant, which is already used in agriculture in many areas anyway, right?

Stefan Kopfinger: Exactly. This is used, for example, for the tracking of tractors. RTK-GNSS has an additional correction signal, with which one is instead of in the meter range in the smaller centimeter range. Normally, I then have a program with which I can do the track planning. With that, you can tell the robot in advance what it has to do and you can also monitor it normally through all the sensors. I can see the position on my computer or smartphone, and I also have a bit of remote control. That is the basic logic, where I tell him what he has to do and also check this.
On the robot itself there are often certain safety sensors, usually a touch sensor - a tactile sensor - around the outside, because otherwise he would drive somewhere against it. In addition, it often has a LIDAR system ("Light Detection and Ranging"), which checks in the greater distance to the front, whether there is anything in the way. This is also known, for example, from the brake assistants in cars. These brake a little in advance as soon as people are detected in the path. And this happens before a tactile sensor actually triggers a stop and then brings the machine to a complete halt.

What challenges does digitization bring, especially in rural areas, e.g. the keyword "mobile phone coverage"?

Christian Metz: The point is that I, as a farmer, can at least make phone calls out in the field. But it often fails already and the reception is not always good. But since we work with GNSS/ RTK technology here, we have to be pretty accurate and for that we also need very good network coverage . So network coverage, the performance of the network in rural areas, is a big issue to be able to use these devices.
If you go one step further, in the future we might even want machines to communicate with each other. If instead of one big tractor, I have three small robots that agree: "I'll drive on the left, you drive on the right, but then we shouldn't meet and collide later at the end of the field." So collusion is an issue there, too.
And the assumption is that robots like this might not be cheap. If I purchased this equipment, then of course it has to be serviced. I.e. the whole service structure behind it must be developed. The companies must be approachable accordingly and the device must also be easy to operate for the farmer. The part has to work so that you can work efficiently. And that is perhaps then also such a point on the subject of acceptance, how these devices then also come into practical use.

Who actually manufactures such robots? The major agricultural machinery manufacturers? Who is active in the market?

Christian Metz: The manufacturers are getting on board and there are also differently sized robots. We had it with the RTK signal, there it is also about autonomously driving machines. So not only the robot that looks abstract, but also the tractor that simply drives autonomously. That's also already a degree of digitization that we have.
The technologies are largely being developed by start-ups , both from agriculture and from other industries. The whole thing is very interdisciplinary and the interest is very high. Stefan, you also have projects running where start-ups in Ruhstorf in Lower Bavaria are doing research on this - and internationally, right?

Stefan Kopfinger: Exactly. Most of the robots available on the market are actually from start-ups. But the large manufacturing companies, both for tractors and for agricultural technology, are also building prototypes or pilot series. In addition, many universities have research projects where robots are built in student teams, as well as tested and improved in the field. What is currently available for purchase, however, is rather from young companies.

Can I imagine it so that with you two exactly that comes together, so these different competencies to bundle?

Stefan Kopfinger: As a research institute, we participate of course again and again in joint research projects, in which always also industry partners and partners from the university are involved. There, we definitely try to bring together the important players in order to drive the whole thing forward.

Christian Metz: This is exactly where our competence network comes in. We unite the players in agriculture on these innovative topics and also network and combine them with other industries. And there the automotive industry is a very nice example. In the automotive sector, we often offer exchange events or events with us with the various sectors. And that's where agricultural equipment manufacturers then exchange ideas with automotive manufacturers as well as mobility providers and benefit from each other, for example, on the topics of sensor technology, complex landscapes, complex environments in agriculture and the environment in the city where many people move around. To use farming lingo, in the competence network , the ideas fall on fertile ground.

How does it work in practice as an agricultural business? Do I buy or rent such a device as a robot?

Stefan Kopfinger: Most devices at the moment are devices that I buy. This is also with the texture. I'm talking here now mostly about the FarmDroid FD20, this sows and chops according to the same predefined grid. This works because the robot places the seed so precisely that it is where it expects it to be. This is a system that I actually have to use throughout the season, and very regularly. To that extent, in this case, it makes more sense to buy it than if I were to get this service over and over again, every week at a very short interval. In addition, the machine is relatively slow on the road, but in return it does this very costly manual work.
However, this is also available as a service. Some manufacturers, such as the start-up Farming revolution, offer the service "hacking" simply with their hacking robot. However, the robot does not know the position beforehand, but works with cameras and on the basis of a trained pattern recognition it distinguishes weeds from weeds.

Does as a service perhaps also increase acceptance? If you can try out a hacking robot once and you see how it replaces manual labor, then perhaps the incentive is more likely to make the investment?

Stefan Kopfinger: Exactly. That is a nice start. You can then see how well the machines work.

Christian Metz: I also think it's just important to see the whole thing, because such a robot, we mentioned the Mars rover at the beginning, is very abstract. In my opinion, they often look like mobile camping tables of all sizes that drive across the field. With 40, 50 or 60 such applications in Bavaria, the probability is not so great that I discover them while cycling and thus get used to the sight.

Now we want to take a look into the future. Stefan, what awaits us there? What are you currently researching at the State Institute for Agriculture?

Stefan Kopfinger: Well, we are currently working on system integration in crop production systems. Say, on the so-called strip intercropping, a small-scale crop production system. We have a new research project since this year: "Crop production system of the future - biodiverse - soil-friendly - digital", where the aim is to examine this small-scale system and its advantages in terms of erosion prevention, biodiversity enhancement and plant health. However, when I become so small-scale, it is of course clear that the work becomes more time-consuming. That's why the fields have grown so far, that's why they've been consolidated a bit. And we hope to find a remedy with the robots, so that this remains somehow economically feasible by carrying out these processing steps autonomously. We are currently determining working hours to see whether everything is as we imagine it.
Basically, it is currently the case that mainly small niches are served, such as sugar beet. There we have just under four percent area share. If we now see the ecological advantages, perhaps of these machines, then the impact would of course be much higher if we were to go for other crops as well. For example, corn, which simply has a much higher area share of 20 percent, would have a greater impact. On the other hand, this attractive niche will perhaps be saturated at some point. But when that time comes, the machines will also be more capable and less expensive, and they will also make sense for crops like corn, where you can't save so much money by using a chopping robot, but where it is still profitable to use such technology.

Christian Metz: That's an interesting approach. So if you approach it abstractly now, you can say, I'll grow exactly the fruit in each place that grows best there, and best of all the fruit next to it that can mutually benefit from each other. Stefan, is that the abstract thought behind it?

Stefan Kopfinger: Yes, that is even a second new approach that you are addressing here, Christian. We are not quite so small-scale now. We have 15 meter wide strips and of course there are visions where really one plant grows next to the other and everything is very small-scale and each plant is individual. That's a very exciting idea. It might not come in the next three years, but it might be possible at some point.

I think there's also a lot happening in the area of digitization in the future. What's coming up, Christian? Which technologies do you want to implement with?

Christian Metz: A technology that is often used is the drone. We are now also taking to the air in agriculture. Not only in South America it is common, if you have the images of aircraft there in front of your eyes, which fly over the areas, but also with us in Bavaria it comes more and more. With the drone, I can also monitor the field from above. I can also selectively apply beneficial insects, i.e. carry out biological pest control. I know then, there on the field are my pests, which I do not want and bring out there then the beneficials, which eat the pests, completely purposefully from above.
And with such drones I can operate e.g. also animal protection. Fawns are deposited by their mothers preferentially in the high grass. Before the meadows are mowed, the situation can be analyzed with the help of drone flights and the animals can be specifically taken out of the areas before mowing. The applications of drones can combine animal welfare aspects with forestry, agriculture and society. Another point of new technologies is also bark beetle detection. You can see from the examples, there are quite a few industries that work together and use the same technologies and that supposedly have the same use cases.

The interview was conducted by Christoph Raithel, Team Leader Event at Bayern Innovativ GmbH.

Listen to the full interview as a podcast: