Digital agriculture: connected farms and farmers

Logged in is clearly the watchword for this project. Irstea teams have been following trends and developing systems capable of closely monitoring fields in real time. The aim is to collect data that can help improve farming practices from a production and environmental perspective. "In the agricultural world of the future, everything from fields and machinery to buildings will be connected. These connections will be used to transmit data and new services will have to be invented," comments Jean-Pierre Chanet, researcher at Irstea's Clermont-Ferrand center and President of the French Association for Artificial Intelligence (AFIA).

Using a Wi-Fi network to anticipate agricultural risks

Farming involves many risks to crops, including water stress in plants, pests, risk of lodging, etc. Early detection is crucial to deal with these phenomena.

Currently, farmers can make use of weather stations or sensors embedded in their machines to monitor parameters such as the vegetation index, output, and the water status of crops and the soil. However, this data can only be collected at long intervals. Satellites can also provide this information, but with low spatial resolution.

Wireless sensor networks (WSNs) are a promising technology currently finding its niche as a way of complementing existing solutions and making up for their weaknesses: Implanted within plots, they can ensure continuous monitoring of various parameters. "This is a new generation of onboard systems linked to wireless communication technologies. Using these devices, we can collect, store, process and transmit data," explains Gil de Sousa, researcher at the Irstea Clermont-Ferrand center.

The CROCUS project is working to improve sensors capable of measuring the status of plant cover. "This can be characterized using two parameters. The first is reflectance, which is the proportion of light reflected by a material, which in this instance is plants. This gives an indication of the size of coverage, the age of the plants and the status of nitrogen nutrition. The second parameter studied is transmittance, which is the amount of light that passes through the plants. This is an important indicator for the structure of the cover," describes the researcher.

What does this new sensor look like? It is a very small computer that includes:

  • a micro-controller (miniature integrated circuit or sensor, such as a rain gauge or a thermometer)
  • a connection interface (USB port)
  • a wireless communication module, such as Bluetooth or Wi-Fi.

Some devices already on the market contain up to sixty sensors! These measure CO2 levels, soil humidity, temperature, radiation and luminosity and even provide vehicle detection.

More importantly, sensors never work alone. They are permanently communicating with all other identical sensors distributed across one or several plots. To understand, we head to Champagne-Ardenne, where the project is underway. Here, experiments are running to monitor vine irrigation and crop development. On plots, wireless sensors have been set out in strategic locations that are close enough to each other to provide the closest account of real conditions encountered in the field. It's as if the sensors were talking to one another.


Each sensor sends the collected data to a central collection center. This is the brains of the WSN that combines all the data collected across the plots and makes them usable. "It’s a complete information system that stretches from field to farmer. The final aim is to be able to use the data to help farmers deal with any potential risks that could occur on their plots," explains the researcher. Once processed by the integration platform, the information is transmitted to users over the internet. It is then used to populate specific applications used to predict the risk of pest outbreaks, update the risk of lodging, predict output, etc. When linked to decision-making support tools, it can provide important recommendations to farmers and technicians. "The project has significant potential for value creation," concludes the researcher. "Based on wireless sensors, the proposed monitoring system has significant development potential due to the information collected (water status and ground temperature, plant status, climate conditions, etc.) and its application conditions, particularly for field crops."

Drones and robots for precision agriculture

Other promising technologies are also being studied alongside the wireless sensor networks. The guiding principle for some of these is to aim high! This is the function of image sensors loaded onto drones, or small automated flying robots. The Rhea project, started in 2010 by the Irstea Montpellier center, is focused on developing their use as part of the chemical and physical processing of weeds or any unwanted plants. These small devices can be used to fly over plots at low altitudes and capture images in order to define areas requiring treatment. At the same time, other, ground-based robots are sent into the plots to ensure the areas identified are treated.


> Data sheet/ Drones: land seen from the sky


"Drones and equivalent land-based robots enable precision agriculture. Currently, when weedkillers are applied to plots, 80% of the product is useless as it is applied in areas that do not require treatment. Not only does this damage the environment, but it is also very expensive! By undertaking precision weeding, weedkiller can be applied only to areas that need it," explains Gilles Rabatel, researcher at the Irstea Montpellier center.

In Europe, a large number of agricultural products could be affected, such as wide row crops (field tomatoes, corn, strawberries, sunflowers, cotton), narrow row crops (wheat and winter barley) and tree crops (walnuts, almonds, olives). Even open forests could benefit from this technology. "The aim of the project is to reduce the use of agricultural chemical inputs by around 75%. As a result, product quality will be improved along with the health and safety of people working on the plots. Rhea is a more sustainable way of managing crops," explains Gilles Rabatel.

"Currently, we are still limited by the fact that detecting weeds using images is still difficult. There are many weeds, and sometimes they can be difficult to differentiate from crops. At the moment, we know how to separate crop rows from the areas separating the rows. Anything that grows between the rows is weeded," explains the researcher. We expect progress to be made with more complex image sensors that will make it possible to provide a better understanding of each plant's spectral signature, the radiation reflected by a surface, according to the length of wave received. Each type of plant or tree has a unique spectral signature, depending on its intrinsic features. Research is also underway to detect orchard illnesses, also using spectral signatures as they change with disease. Stay tuned for more news on this project!