FIRA19 Live notes

Davide Rizzo
FutureAg
Published in
7 min readDec 10, 2019

10–11 December 2019, Toulouse France

Live notes from the 4th International Forum of Agricultural Robotics. I’ll update this page along the two days.

Ethical and societal questions on agricultural robotics

The opening conference by Alexei Grinbaum, researcher at CEA-Saclay.

One key point: learning without understanding. That means having less words than collected and processed parameters.

Robots serving field crops

Agrointelli and John Deere debate with Olivier and Maurice, two farmers from the region.

Main point: make it simple, because farmers need information to decide more than data to analyze

Robots serving fruits and vegetables.

Flexibility for adaptation is the key point in the interaction between farmers and engineers. The second FIRA round parallel table includes Dino from Naïo (lettuces weeding), SYHA (tomato harvesting) and Hortifruit (blueberries spraying and harvesting).

José Miguel Arizabalo Barra, corporate CIO Horticulture S.A. (Chile), stresses the need that robots have to be a service. This implies lifelong support from the manufacturers, even if they are far away from the farmers that use robots. Continuous adaptation to each specific farm is crucial for the reliability of this new equipment.

In the view of Jean Inderchit, DINO product manager Naïo Technologies (France), the value creation and value chain structure are still unclear because the available robots are still under development. Early adopters are still looking for different models to find the most suited to their particular operating system.

However, the uncertainty in the value chain creation harness investments because of the difficulties in calculating the ROI. In particular, it is critical to define how to improve collaboration to scale business cases and cost and gain evaluation.

How robotics change job and support agriculture?

Robotics for breeding

One out of two breeders that will take up a dairy farm will buy a milking robot. This choice is even considered the only way to achieve economic viability.

Feedback from a dairy farmer and a support agent highlight relevant points. The automation facilitates the farm take up, especially the family by easing the job charge in the transition phase, namely for the ageing parents.

Automation is part of the deep farm restructuring process, both on animal breeding and management and on production quality. In perspective, this impact also the feed cropping system.

Robot setting offers huge opportunities, yet too many parameters can be defined. So extension and support services are required to personalise factory settings. This step is crucial for the good success of robot exploitation.

Once the settings have been suited to the specific farm and farmer’s needs, support services need to frame the good management. As an example, a water temperature sensor could be harnessed by limestone water. External monitoring could help so to address practical issues.

Finally, farmers can be supported in exploiting data collected. In this sense, good maintenance increase robot operating lifetime, so data collection.

In the end, the milking robot allows for greater flexibility in livestock management. Though, is crucial to respect the ROI and financial integration of the robot costs. As so? the support service should check if the farmer is robot-ready and capable to deal with different management.

Altogether 80% of dairy farmers are satisfied with his/her farm if having a robot, against the 65% without a robot (according to a local survey).

Robotics for vegetables

Matrhoeu Follet, an early adopter of the Naïo weeding robot witness of his experience. Of notice, the buying of the robot was supported within the conversion to organic farming by a partnership with consumers and final users.

In this vein, the robot was particularly appreciated because of the lighter weight (thanks to electrification) that reduces soil compaction even in the case of multiple tillage operations.

Marhilde Ceaux, from Naïo Technologies, os on charge of the after sale services. The robot adoption includes three parties: the manufacturer, the robot and the farmer. First of all, the farm is checked against the hard technical robot constraints. Both the farm and the farmers have to be robot-ready because some adaptation could be required to take into account the robot technical evolution. In this vein, the personalized support for the farmer is a win-win situation also for the manufacturer, who can learn from the field operationalization and so improve the machine for multiple use cases.

The first advantage of robot adoption is greater comfort, even though a collaboration between the farmer and the robot is needed, for instance for in-row weeding.

On the one hand, replacing a fuel (old smoking) tractor with a silent electric robot allows for a quieter and more pleasant work-frame. On the other hand, the robot requires much more anticipation and foreseeing: it is a preventive tool rather than a curative one.

Education and training are even more needed in horticulture because they have to manage tens of different crops and varieties, on top of which adding robot management skills. In perspective, the farmer hopes that the better working conditions will facilitate the back to the farming of new operators and farmers.

A cliché from the round table about job and support service changes in agriculture in the transition to robotics (photo: D. Rizzo, FIRA 10 Dec. 2019, Toulouse)

In summary, the transition towards robotics is a multi-actor adventure.

Does robotisation contribute to the quality of life at work?

Aline Dronne (sociologist, Aract Grand Est) proposes a focus from a sociological perspective, especially on dairy farming.

Robotisation is a heavy investment. As so, it requires motivation. In addition, it implies a deep change in the job, even though not completely. Surveys showed that the breeders consider that their job is not changed, but even improved. Some of them affirm that the robot allowed for a better knowledge of the herd as a whole and of each animal. Nevertheless, the farmer should be ready to manage data to really exploit this opportunity.

The milking robot could potentially enable a better balance between personal and professional aspects of life.

New technologies risk nevertheless to deprive farmers of the equipment mastery. On top of that, there is the risk of isolation because the community built on traditional tools and knowledge is destructured by the new machines that are for few people.

In conclusion, a hidden risk of robots and new technologies is the constant connection and a never-ending mental load.

The perspective of health services

Simulating driving conditions, reducing load movement as well as the contact between operators and phytochemical are some examples of risk prevention.

New technologies are therefore a possible mean to prevent and reduce risks. In this vein, virtual reality could be used to increase risk awareness.

Insurance: should we ensure agricultural robots?

First, we should define what is a robot, then looking at what is the regulation.

Mobile robots moving on the road are submitted to road regulation. However, even immobile robots as milking robots could be ensured because of their value.

In general, robots are quite secure because of their multiple sensors.

Robots can be insured at multiple levels: environment (people moving around), the operator and the robot itself.

In perspective, some issues can be raised about agricultural robotics insurance: cybersecurity, steal, changes and uncertainties in regulation, social acceptability and risks of vandalism.

Insurer perspective on agricultural robots (photo: D. Rizzo, FIRA 10 Dec. 2019, Toulouse).

Groupama insured tractors from the beginning, and around 1900. Now they have started to insure both mobile and stable robots, even though for different reasons. They have considered mobile robot insurance to answer Naïo solicitations on this topic.

In perspective, in my perspective, insurance operators should also address the whole digital environment composed of various connected devices and telemetrics.

By the way, agricultural robots for what?

Research and development and uphill actors in the development chain can play a role in scaling up agricultural robots.

Anne Paulhe-Massol, Innovation director at Arterris (France), is considering the provision of robot as a service for neighbouring farmers. The service approach could represent a relevant alternative against the buying of this equipment, especially for small farms, and to overcome affordability issues. In this vein, intermediate actors could act as contractors by buying robots for leasing.

Downhill stakeholders in the agrifood chain should be more involved in agricultural robot development. The goal is to allow them transcribing the quality expectation and the environmental labelling.

Pablo Quijida, R&D portfolio manager at Bayer, opens on the development of varieties more suited for robotics, for instance of tomato varieties more easy to automatic harvesting.

Roland Lenain introduces the workshop by addressing the main scientific locks and wider expectations.

Societal questions identified by Roland Lenain for the RobAgri association (photo: D.Rizzo, 11 Dec. 2019, Toulouse)

Many robots are already around, yet some scientific locks and societal expectations to be addressed. In this context, 70 actors of the French and European sector of agricultural robotics joined to create the RobAgri association. This association is providing a networking framework to boost the development of new tools for agriculture.

Robotic vision applied to weed control

Feras Dayoub from Robotic Vision proposed a very wide and relevant overview of the application of machine learning for computer vision.

Classical accuracy measures to train models have to be reviewed based on the sensor performances and the operational conditions. In addition, the environment of deployment in real life is changing on the go. Altogether this can result in model fragility, which is a failure in its labelling performances.

Originally published at http://agronoter.wordpress.com on December 10, 2019.

--

--

Davide Rizzo
FutureAg
Editor for

Agronomist doing research & teaching in (geographic) data management to understand agtech opportunities for farmers https://twitter.com/dav_rizzo