In June, it was announced team AuTomatoes won the Autonomous Greenhouse Challenge. As a member of team AuTomatoes, Hoogendoorn Growth Management spend half a year growing cherry tomatoes remotely without entering the greenhouse. With the use of Artificial Intelligence (AI) the team scored best on all aspects, including yield and sustainability, beating the other 5 teams and the professional tomato growers. We sat down with Silke Hemming, Head Scientific Research of Team Greenhouse Technology at Wageningen University & Research (WUR) to consult her about her 15 year-long journey towards autonomous cultivation. Hemming is considered as the driving force behind the competition.
A 15-year long journey
The WUR is already working on models, sensors, intelligence etc. for a long time… And although you may think autonomous growing is completely new, the WUR has grown their first autonomous crop 15 years ago. The cultivation of sweet peppers was automated by a mechanical greenhouse climate model. Hemming: “I was always fascinated by the work of my colleagues and never understood why there wasn’t a follow-up. Why it wasn’t brought to practice on a larger scale. But we couldn’t get any funding for this.” That changed two years ago, when Hemming met David Wallerstein from Tencent. “Tencent wanted to cooperate with the WUR. Among other colleagues I shared my ideas. Tencent focusses on intelligence, WUR focusses on horticulture. A match made in heaven.” Subsequently, the Autonomous Greenhouse Challenge was born.
From having drinks to merging teams
According to Hemming, “it isn’t all about winning the competition. It is about bringing expertise together, sharing knowledge and creating connections to grow. That’s why the first phase of the competition included a Hackathon.” During 24 hours, 21 teams competed to maximize net profit of a virtually grown tomato crop in a gaming environment. “We wanted the five best performing teams, but we also wanted to scout internationally. Which parties are there, and who thinks this is interesting and exciting?” The Hackathon, organized at the WUR in Bleiswijk, was the way to select those five teams and to bring all these teams from all over the world in touch with each other.
Hemming: “It was so different this year then during the first competition. This year, participants seemed to take the competition more serious. People really felt they were in a competition to win. Last year, there was more interaction between the teams, especially during the Hackathon. There were more discussions and people were having drinks together. They had fun. Some teams had more AI knowledge, others more horticultural knowledge. They realized that and enjoyed the discussions, leading to unexpected presentations for the jury the next morning. Teams were merging!”
Bringing out-of-the-box ideas into practice
The competition allows teams to test their algorithms and improve them. Hemming outlines: “Once I was visiting the Hoogendoorn headquarters, for a completely unrelated matter, and I had the possibility to view their control room. From there, they operated the greenhouse. One of the team members, Evripidis Papadopoulos, said they started to try completely new ways to control the climate and irrigation. Just some out-of-the-box ideas they brought into practice. That is what the challenge is all about. There is a risk to go too far, but if you never try something new, you can’t win.”
According to Hemming, a strong element in the strategy of Team AuTomatoes was Plant Empowerment. The team centered all their actions around the plant. “That is what growing a crop is about. When you find a balance between the growth factors your crop performs well. And up until today, a well performing plant has always brought more income then costs for the resources you put in.” The ratio between costs and yields is definitely important, explains Hemming. Some of the teams really focused on the cost perspective at the start by economical use of e.g. the heating and CO₂ supply. “If you choose such strategy, you want to save resources and costs. Eventually they realize it may have negative consequence for crop yields.”
Toasting to the future of autonomous growing
After a half year of growing tomatoes autonomously the results were announced, causing champagne bottles to pop for team AuTomatoes. “The results were really close. What we saw is fantastic, the potential of remote control, algorithms and AI had been demonstrated. This provides perspective for the future. It shows autonomous growing is possible.” Despite this great outcome, there are still steps to be made. The human is still very much involved. But the route to autonomous growing has certainly started!
And now the big question everyone is wondering: will there be a third edition of the competition? Hemming: “I really hope so!”