Which team will develop the best algorithm to allow the greenhouse to grow dwarf tomatoes autonomously? This question is at the heart of this year’s Autonomous Greenhouse Challenge, a competition organized by the Greenhouse Horticulture Business Unit at Wageningen University and Research (WUR). After the first part of the challenge, five more teams will compete for the win. The team that has achieved the most profitable harvest of dwarf tomatoes using its algorithm by mid-December is the winner. Throughout the summer, the teams are building their algorithm that will control the greenhouse from September onwards.
Growing crops in a greenhouse without human intervention. It sounds like a utopia, but according to Steve Maree, a data researcher at WUR Greenhouse Horticulture, this picture is not so unrealistic. “Greenhouses are becoming increasingly ‘smarter’. For example, technologies ensure that water and nutrients are managed at the right time and at the optimum temperature and humidity. Data collection also increasingly enables farmers to make better decisions. This leads to more efficient crop production. This is particularly important given challenges such as increasing food demand, rising energy prices, reducing CO2 emissions and staff shortages.
According to Mary, WUR is one of the world’s leading companies in autonomous farming. “Within WUR, a large team of researchers is involved in different aspects of autonomous greenhouse farming, from design to operation. One of the major projects we have worked on in recent years is AGROS. Here we have shown that we have the ability to control cucumber cultivation completely autonomously. Growing based on AI models and expert-based models has led to results similar to those achieved by farmers when they controlled the greenhouse themselves. Proving practical feasibility is a huge achievement in autonomous farming.”
Independent Greenhouse Challenge
In addition to its own research, WUR also tries to contribute to the development of autonomous greenhouses in other ways. For example, through the Autonomous Greenhouse Challenge. Marie explains the goal of this competition. “With the challenge, we want to bring together people with different expertise from all over the world. From plant experts to technology experts. In addition, it offers participants a unique opportunity to gain practical experience in autonomous agriculture. Testing technology in greenhouses is expensive. By making our research greenhouses available, students, researchers and companies can test their devices for free and in a fun way.”
Growing dwarf tomatoes
The Independent Greenhouse Challenge is taking place for the fourth time this year. Every year there is a different crop as a basis, says Marie. “This year we have chosen dwarf tomatoes. Unlike the plants in previous editions, they are grown in pots and sold as a whole plant. This requires new options, because there is no single independent system for each crop or greenhouse. The challenge is to sell the plant with ripe and semi-ripe tomatoes, so that consumers can see the last tomatoes turning red at home. We try to expand the challenge every year and make it more difficult, for example by providing less and less space for manual production. We do this because greenhouses are becoming increasingly independent in practice.
Part one is done
23 teams from all over the world are taking part in the challenge, the first part of which was completed in the spring. The teams had to complete various tasks. Marie: “The first task was to develop an algorithm or model with which they could estimate the condition of a dwarf tomato plant based on a photo. They also had to grow dwarf tomatoes virtually for two months. The final stage took place in June on 6 and 7 Plesvik, where our research greenhouses are located. The teams had to develop an algorithm in a hackathon to identify whiteflies in yellow sticky traps, which is why it is important for farmers to keep this pest under control.”
Five teams compete for the win.
The first part of the challenge ultimately pitted five teams against each other for the win. These five teams had to develop an algorithm that could control the greenhouse completely autonomously, says Mari. “Each team gets their own greenhouse where they have to grow and eventually harvest dwarf tomatoes. In their algorithm, they can manage different knobs, such as water release, temperature, light, and humidity. The trick is to find the optimal one here. More light and heat means you can harvest earlier, but it also costs more energy.” The greenhouses are equipped with standard sensors, but teams can also install additional sensors.
Algorithm development
The teams worked all summer to develop their algorithm. They use the WUR simulator for this purpose. Marie: “This simulator is a digital version of the tomatoes that will be grown. We fed the simulator data from a previous experiment. We grew dwarf tomatoes in the same greenhouse, but by hand. This gives us a better idea of what parameters are important for crop development and what sensors are needed for this. The team will build their entire algorithm based on this simulation.”
The most profitable harvest wins.
The algorithm should finally be ready on September 2, says Mari: “From that moment on, the greenhouse will have to do the rest of the work. The tomatoes must be harvested by December 16 at the latest. The winner is the team with the highest net profit based on the price per pot, the number of pots worth selling and the costs incurred. We set the price per pot at 80 euros in advance. If there are less than average red fruits, there are fewer, if there are more red fruits, the taste also affects the price. We base this on the dry matter content, the more the tomatoes taste.
Valuable knowledge and insights
In addition to eternal fame, the winner will also gain valuable knowledge from the challenge, says Mari in conclusion. “They can use their algorithm for further research or for commercial use. Many successful companies have already emerged from previous editions. The challenge will also provide us with new insights. For example, whether the new sensors provide reliable data. But also which strategies will be applied? The most efficient because this is the ultimate goal of autonomous agriculture. As organizers, we do not see exactly how the algorithms work, but we see the consequences of the choices. This provides us with a lot of valuable data that we can use for further research and will help the sector on the way to fully autonomous greenhouses.”
The 4th Autonomous Greenhouse Challenge is organized by Wageningen University & Research, BU Greenhouse Horticulture and Flower Bulbs and sponsored by Tencent, Biobest, Fluence, Lensli Substrates, Pöppelmann, Quantified, Certhon, Vreugdenhil, Gebr. Geers BV and LetsGrow.
source: And
“Total coffee specialist. Hardcore reader. Incurable music scholar. Web guru. Freelance troublemaker. Problem solver. Travel trailblazer.”
More Stories
Brabanders are concerned about climate change.
The “term-linked contract” saves space on the electricity grid.
The oystercatcher, the “unlucky national bird,” is increasingly breeding on rooftops.