Roland Lenin: “Robotics will be one of the levers of agriculture”

Is agricultural robotics becoming a reality in France?

The answer is multiple. First things first, agricultural robotics is already a reality in animal husbandry. Today, the feeding robot is largely democratic, and there are also robots for feeding, animal feed … It’s growing for a simple reason: it’s a structured environment, where robots “feel good.” Robotics develops more easily when you have the environment, the state of localization, when you place it on a closed, clean, marked site. All this explains the developmental differences between animal husbandry robotics and field robotics.

How do we explain that there are no more robots in the field today?

For robots to be effective in open terrain, locks still need to be lifted. They must act according to their environment and the work assigned to them. They need to be able to recognize dangerous situations and adapt their behavior accordingly. That’s the decent thing to do, and it should end there. For example, ignore a branch stuck in a row of vines, but stop if one person is sitting in the field. It’s quite complex: you need the right sensors, color, geometry and to work in any light condition – including night … there are a number of issues that work.


Can Artificial Intelligence Help Overcome This Identification Problem?

Regarding AI, we are constantly making progress. Computing power allows us to learn far more than we have in the past, and to learn better. But we have a problem with AI: the “black box” aspect of neural network technology. In short, we cannot predict the response to an unprecedented situation outside of the AI ​​learning base. This makes the robots equipped with them difficult to verify and certify.

What methods are currently being explored for robots in plant farming?

The first track is conventional agricultural automation. A simple example is a self-guided tractor. That way, we don’t imagine new ways of producing, we just make a job less painful. The second track is to tell yourself: a robot can provide services when people are doing something else. Since it can be dangerous, but can also affect the ground, to make a big robot work independently, we will work on small machines. Thus, we see the appearance of weed robots, such as Oz or Anatis. In my opinion, instead of having big complex machines, in the future we will have smaller, more basic, less energy consuming machines and we will adjust the size of the fleet according to what we want to do.

One of the arguments of robotics is to help agriculture in greening. Can you explain to us what he brought to the field?

It’s simple: alternatives to productive models derived from mechanization and chemistry are new methods such as organic farming, no cultivation, crop rotation, which requires intervention in a way that makes the field more frequent and much more. Precise and local way. So it is more humane time, repetitive and painful, which creates a labor problem, there are very frequent muscle disorders among the workers. Robots, for their part, do not require human time and are able to intervene in a precise and repetitive way: they are truly profitable in agriculture, energy and more profitable inputs. This is why I am convinced that robotics is one of the levers of agriculture.

INRAE ​​robot


What are the main obstacles to its development today?

We mentioned that in addition to the technical difficulties, the brakes are also regulatory. Today agricultural equipment, including robots, is subject to “machine guidelines”, which are adapted to the manual operation of the machine. There should always be someone under permanent control, under the control of the device, or under remote supervision. This greatly limits the interest of a robot … In terms of research, we will proceed with a 100% guarantee of the robot’s behavior: this is a big challenge, which involves significantly artificial intelligence. There are also other legal hurdles, such as a ban on crossing a rolling robot, even occasionally, a road between two plots. It complicates development and testing, not to mention commercialization. However, we are working to move this legislation forward.

What will be the reasons for accelerating agricultural robotics in the future?

In the long run, I strongly believe in sharing basic technology, databases, development infrastructure. We must add the cost of research and data for the benefit of both private players and public institutes like INRAE. This will be a real cause of acceleration, which the community is now promoting within the RobAgri Association. There is also the question of standards: just as we have a highway code for vehicles, we will need a “common dictionary” standard shared by agricultural robot manufacturers to facilitate their operation by farmers.

Leave a Comment