Thomas Watteyne: "In Argentina alone, connected agriculture could have saved 10,000 jobs in one year"
Sensors - © Inria / Photo C. Morel.
Thomas Watteyne is a member of the Eva team at Inria Paris. In 2016, as part of the "Save the peaches" project, he equipped Argentinian orchards with sensors in order to predict the arrival of frost and save the crops. The team is now planning to use the same technology in order to predict snow melt in the Sierra Nevada.
> With the Eva team, you have taken part in the "Save the peaches" project: what is it about?
Thomas Watteyne:"Save the peaches" is a connected agriculture experiment where we use the Internet of Things in order to predict episodes of frost in peach orchards. These episodes pose enormous problems for farmers, and I was surprised to learn just to what extent. The vast majority of fruit trees go through the same critical period: the spring. When the flowers open, all you need is a temperature difference of 3 or 4 degrees for several hours and they freeze and fall, preventing the fruit from growing. In 2013, in the Mendoza region in the west of Argentina, almost 85% of fruit production was lost, just because of a few hours of frost. This represents around 10,000 jobs and 100 million dollars!
> Can the farmers do anything to save their crop?
Yes, and they are already doing so. If they think that there is a threat of frost, they install radiators in the orchards, or they use big fans - even helicopters - to make the air circulate and heat up the atmosphere. Dealing with episodes of frost is not a problem for the fruit producers - the problem is knowing when they will arrive.
> Isn't it sufficient to monitor the weather?
Of course, monitoring the weather forecasts can help. However, they really are not precise enough to work at orchard level. So the farmers equip themselves with extremely sophisticated personal weather stations, installed directly in their orchard. But even that isn't enough! A weather station cannot pick up climatic nano-phenomena that may arise within the orchard itself. Even the wind direction can alter the temperature within an orchard by several degrees, between the crown and the top of a tree, or the middle and sides of the plot of land.
> So that's where you come in!
Exactly. Rather than using a weather station, we use a large number of sensors placed in the orchard, from which we gather measurements every 30 seconds. This give us a full representation of the situation of the orchard, in real time.
> What do these sensors measure?
They measure the temperature and humidity of the air, at four different levels on each tree. In a few months time, we will also measure soil hydration and temperature.
> How does this data help to predict an episode of frost?
We are developing "machine learning" type tools in order to analyse all of the data gathered by the sensors and to know which conditions lead to an episode of frost. Once we know these combinations of events, then the system will be able to predict them. The farmer will then receive a text message telling him "there is an 87% chance of frost occurring in the southern part of your orchard between 1 am and 4 am." If our system had existed in 2013, we would have saved 10,000 jobs and 100,000 million dollars!
> This is indeed very impressive. What has been the reaction of the producers with whom you have worked?
They are extremely enthusiastic. As a result, for the last six months we have deployed a prototype in the Mendoza region in Argentina. We are working with the Argentinian Institute of Agronomy, which is directly linked to all of the country's farmers. We are delighted to work together with them and to be able to help to create a direct impact on the life of the producers.
> In concrete terms, how do these sensors work?
Everything works wirelessly! We are working with cutting-edge technology, designed by the company Linear Technology/Dust Networks and in partnership with Metronome Systems. They provide us with wireless systems that we connect to the sensors. As a result, there is not a single wire in the orchard. Our sensors are also very energy-efficient. They only require two simple AA batteries for 10 years of use!
> Can these sensors be used in other fields?
Yes, we are working on that all of the time. Our team has just got back from a month-long trip to the University of Berkeley in California. We worked on the deployment of a network of sensors based on exactly the same technology. We use the same information transmission systems, the same tools to connect the sensors and configure the deployment. But in this case the sensors are measuring snow melt in the Sierra Nevada mountains. The hydrologists use this data to better understand and model snow cover and, for example, predict the output of a hydroelectric dam. The project is called "SnowHow". We use ultrasound range finders to measure the quantity of snow remaining in certain strategic areas of the Sierras. We have deployed 945 sensors, that emit every 15 minutes. We believe it is the biggest wireless environmental monitoring network in the world, and we are very proud of it.
> So the Internet of Things can change all of our lives?
Yes, and we are constantly asking ourselves how. For the time being, society's approach towards the Internet of Things is very much aimed at the consumer, with the development of connected coffee machines, etc. However this technology can be used in many other areas - in agriculture, as we have seen, but not only. My next research project with the Eva team also involves the Internet of Things, and it is very exciting. We are going to work on equipping one of the biggest marinas in the Mediterranean with sensors in order to follow, in real time, where the boats are moored. After smart cities and smart parking - the smart marina! Ziran Zhang, postdoctoral researcher, will work with me on this project as of December. For my part, I am developing the intrinsically safe Internet of Things concept. This means developing highly-reliable and highly-secure technology, for which we can predict the communication latency. This enables us to use these technologies in industrial control loops. Obviously the development of such technology poses a large number of safety and security questions (can it be hacked?). We are working on these issues with the IETF 6TiSCH standardisation group and hope that we will soon be able to test it. In my research, I am fortunate to be surrounded by a great team. It is a joy to work at Inria. I am supervising 3 postdocs, 2 thesis students, and we operate a little bit like a start-up. With them, I am in heaven!
This work is done in collaboration with Prof. Dujovne's team from Diago Portales University in Chile and Prof. Gustavo Mercado's team from Nacional Technologica University of Mendoza in Argentina, as part of a SticAmsud collaboration.