Robotics in the Orchard: Between Co-development and Operational Comfort

Léger SAS and Maison Martinet: Three Years of Co-development in Hazelnut Growing

Mathieu Martinet - Léger SAS users, grows 40 hectares of hazelnut trees in Lot-et-Garonne (south of France), a still young plantation of which only half has reached productive stage, alongside a processing facility that handles his own harvest and that of partner producers. A hybrid profile, at the boundary between food processing and farming, which creates very particular organisational constraints.

“A good worker who can handle multiple tasks and who we want to keep, we'll pay them almost the same as a manager, between €35,000 and €40,000 a year, that's the reality,” he states upfront. This economic observation is at the heart of his thinking on robotics. In hazelnut growing, three operations concentrate the bulk of the labour without directly adding value: mowing, which is regular and essential since harvesting takes place on the ground, managing suckers, the constant regrowth at the base of the bushy hazelnut tree, and spraying, which is ideally done at night for agronomic reasons. “Having an employee working at night is a bit complicated, so it's always me who has to do it.”

Mathieu Martinet has been a beta-tester of the Léger robot for three years. He has not yet taken the step of purchasing it, he is waiting for the productive area of his orchard to reach a sufficient economic balance, but he actively participates in the machine's development. “We drew up the specifications together,” he explains.

Max Rigal, innovation manager at Léger SAS, confirms this co-construction approach: “Mathieu helps us a great deal. We regularly use his orchard to test how the robot and its implements work.”

Four applications are currently being targeted: sucker removal, shredding of pruning residues, inter-row weeding at the base of the row, and spraying. The economic case is already clear for Mathieu Martinet: “An autonomous machine will cost around €150,000 without implements. The maths is straightforward: we'll pay €25,000 in annual instalments.” And beyond the instalments, there is the time freed up, “the time we save on these tasks is time where I'll be making money elsewhere” even if he acknowledges he has not yet factored this into his calculations.

On the evolution of the technology, he is enthusiastic: “Even ten years ago, robotics felt a bit clunky and wasn't seen as something very serious. With what we've been experiencing over the past two years, particularly with AI, I think we're at the very first step of a revolution in agriculture.”

The next steps for Léger SAS: continuing to adapt the implements but also expanding to other crops. “We are also working with a yellow kiwi producer who is waiting for the machine to be adapted to the specificities of his crop, which is quite different in terms of orchard configuration and cultivation techniques,” explains Max Rigal.

⇒ Watch the replay here
Watch a demo of the RoboCut360 - videos in French with English subtitles

The GoTrack Solution at Philippe Planavergne's Farm: When the Tractor Becomes Autonomous

Philippe Planavergne runs a 72-hectare single-block farm in Tarn-et-Garonne (south of France), of which 60 hectares are apple orchards, half organic with scab-resistant varieties and the other half conventional. For two seasons now, he has been using the GoTrack solution, a retrofit kit developed by a Polish company and distributed in France by Vantage, which makes an existing tractor semi-autonomous.

“It's not a robot,” he is quick to point out. GoTrack connects to the tractor's electronics and takes control: the lift, speed, acceleration, power take-off, and forward movement. Precision is centimetre scale thanks to RTK GPS. And if the system fails, loss of signal, breakdown, “it remains a tractor I can continue to use completely normally.” This reversibility was for him a non-negotiable condition.

The main use: crop treatments. On 30 hectares of conventional orchards, the intervention window is limited to 3-4 hours in the morning. Two tractors treat in parallel, one driven by an operator and the other running autonomously, allowing 15 hectares each to be covered. “In cases where my employee isn't there, I don't end up treating everything alone all day long. It gives more room to manoeuvre and less stress,” explains Philippe.

Getting to grips with it was not immediate. In the first season, the tractor's route with the GoTrack solution had to be mapped to mirror his own usual driving style, one row in two at 8 km/h. The problem with this configuration was that the tractor stopped too frequently. The route therefore had to be redone, this time at 7 km/h and covering one row in four to leave more room for manoeuvring depending on soil conditions. “Once you get used to the recording process, it's straightforward.”

“The system is packed with safety features”: a front-mounted LIDAR, thermal cameras, automatic stop if the tractor deviates more than 10 cm from its trajectory or if an obstacle is detected, the tractor stops, restarts, re-engages the power take-off, accelerates, and then resumes its task. The trade-off: the orchard must be kept clean. Grass that is too tall triggers the sensors.

Out of 300 hours of annual treatment, 150 hours are carried out autonomously. Philippe retains control during key periods, particularly flowering, “because when the fruit grower is in his tractor, he can see how the flowering is progressing overall.” For the rest of the treatments, the tractor runs without him while he handles administration, “an increasingly important task for the fruit grower.”

The benefits now extend beyond treatments alone. This season, Philippe plans to equip his Frostbuster with GoTrack for frost protection: “You drive around all night, it's really very hard on the employee. Now the tractor will do the job.” The solution also allows modulation maps to be read for chemical thinning and guides the tractor precisely when using a mechanical thinner. “A piece of equipment that brings a whole range of other solutions.”

When asked directly about profitability, the answer is clear: “I say it's profitable. Another point: health and comfort, less time in contact with crop protection products, so it's better for my health and that of my employee. It's also better for my back; I've halved my time in the tractor.” And when asked if he would do it again: “I'd go for it without hesitation.”

⇒ Watch the replay here

Kioti: An Autonomous Transport Robot from South Korea, First Time in Europe

Before getting into the details, Jean-Noël Acquette, Kioti's France manager, offered an interactive demonstration of the Kioti transport robot with Philippe Planavergne, who had been unable to attend the morning demonstration.

The machine, developed in South Korea where around fifty units are already in circulation, is making its European debut. It weighs 380 kg empty and can carry up to 300 kg. Its control modes are deliberately versatile: GPS to define an automated route, a wire cable for intuitive manual guidance, or a remote control. The platform is height-adjustable, making it easier to load from a van or truck. It can also tilt into tipper mode for bulk transport.

Philippe Planavergne was able to get to grips with the machine within a few minutes, guided by the cable. “It handles well. You just need to get used to it,” was his simple comment. Jean-Noël Acquette confirmed that a brief period of training is required, provided by the network of 200 dealers present in France, primarily from the green spaces sector. A larger sibling with double capacity, 600 kg, on tracks, is currently under development.

⇒ Watch the replay here
Discover the RT100 in demo | Vineyards & Orchards

What These Testimonials Say About Fruit Growing

These two cases illustrate different realities, but converge on several points.

Skilled labour is the primary trigger. Whether it is a matter of finding a versatile employee capable of treating at night, or of not being held back by an absence during a narrow agronomic window, it is the pressure on human resources that drives the decision to take the step well before financial returns are considered.

Adaptation is inevitable and takes time. Philippe Planavergne learned this during his first season: the autonomous tractor does not mechanically reproduce the driver's reflexes, routes have to be rethought around it. Mathieu Martinet has spent three years co-developing the robot's specifications with manufacturer Léger SAS, adjusting the implements to the orchard and his practices to the machine. “Sometimes each side has to take a step towards the other” — to echo words heard during the vegetable session — they apply just as well here: in the current phase of agricultural robotics development, collaboration between manufacturer and farmer is essential.

Reversibility and versatility provide reassurance. The tractor, even when fitted with the GoTrack kit, can be driven manually. The Léger implements are also designed to be mounted on a conventional tractor if needed. This versatility is an important psychological condition for both fruit growers, who cannot afford to have their working tools immobilised.

Finally, partial autonomy appears to be the emerging model. Neither Philippe nor Mathieu is looking to automate everything. One wants to retain control over strategic treatments with high agronomic stakes; the other will use the robot for tasks that add no direct value, in order to free up human time where it truly matters. The robot supports — it does not replace.