News
4th edition of Autonomous Greenhouse Challenge is now really underway
The final phase of the 4th Autonomous Greenhouse Challenge will start on September 2. Five international teams will try to grow dwarf tomatoes as well and sustainably as possible using artificial intelligence and machine learning. “The Autonomous Greenhouse Challenge is a great and exciting competition,” says Monique Bijlaard of Wageningen University & Research, BU Greenhouse Horticulture. “And this time the challenge is extra challenging, because dwarf tomatoes are a relatively unknown crop.”
In modern greenhouses, the level of automation for crop production is already high. Climate control and irrigation are generally automated. However, decisions such as the required temperature in the greenhouse, planting distances and harvest moments are still made by the growers themselves. Based on their knowledge and experience, the growers strive for an optimal yield with an efficient use of resources. This process is subjective and labor-intensive, and requires considerable experience.
Feeding city residents
To accelerate the transition to autonomous cultivation, WUR launched the Autonomous Greenhouse Challenge a few years ago. Each edition focuses on a different crop. In this fourth edition, it is the dwarf tomato. Bijlaard: “Dwarf tomatoes are not yet a very large crop: they are actually predominantly bought by consumers at garden centers. But one of the major global challenges is feeding the residents of megacities. Urban farms can play a role in this. In those nurseries, tomatoes with a high-wire crop cannot be used easily; dwarf tomatoes can.”
The Autonomous Greenhouse Challenge consists of a number of phases. Last spring, teams could participate in an online competition, which concluded with a hackathon in June. A total of 23 teams participated, with more than 200 people. Each team consisted of at least a cultivation expert, an IT expert and a student. The five best teams are now starting the final round. For this final round, dwarf tomatoes are sown and will be planted in pots in the WUR greenhouses in Bleiswijk. Bijlaard: “The teams must actually grow and produce tomatoes. This must be done completely autonomously, with software, without any people at the controls. This means that they control the cultivation with artificial intelligence and machine learning. For this, they can use a cultivation model that WUR has developed for the dwarf tomato. The software must regulate all cultivation measures online, which are then carried out automatically by the climate computer.”
Focus on sustainability
The teams have until mid-December. They must not only achieve the highest possible production, but also work as sustainably as possible. “Sustainability is an important focus: for this edition, we are using pots that have been recycled to a high standard and a renewable growing medium.”
The final result will therefore also include use of, for example, electricity and CO2. Additional actions – such as placing the pots wider – also cost money. The results will ultimately be compared with those of a reference team, consisting of WUR researchers.