Autonomous growing is the next milestone for greenhouse horticulture. To get there, several questions need to be answered. For instance: whether digital twins or AI algorithms can indeed cultivate effectively. And knowing what data are necessary for proper cultivation control. This was explored in the AGROS project, whose first phase was recently completed. Project leader Anja Dieleman at Greenport West-Holland says: "Data often relate to optimizing the greenhouse climate. But it's about the crop."
'AGROS' stands for Evolution to sustainable AGRicultural Operation Systems. The project is a collaboration between Wageningen University & Research and 26 private partners, including Greenport West-Holland. The aim of AGROS is to stimulate agro-technology for ecological and biological functions in greenhouse horticulture, arable farming, and dairy farming.
"Agriculture and horticulture face several major challenges," explains project leader Anja Dieleman. "For example, livestock farming revolves around nitrogen emission, and for greenhouse horticulture, it's about energy use and CO2 emissions. At the same time, many technologies are available that could contribute to these challenges. 'Technology for ecology,' so to speak. In the AGROS project, the possibilities for greenhouse horticulture, arable farming, and dairy farming are being explored, including which sensors and algorithms are needed for this purpose."
Which data are most important for autonomous growing?
"Every entrepreneur knows: you need data to properly steer your business. And so, much effort is put into acquiring data. These data usually relate to optimizing the greenhouse climate. But the greenhouse climate is just a prerequisite for a successful cultivation. So: what data do you really need? For that question, you need to know what you want to steer on. The starting point of AGROS is steering on the crop. But even when it comes to the crop, you can measure a lot, such as sap flows, evaporation, the length of the crop, etc. And you need to know what you want to achieve. We investigated this for cucumber cultivation. The goal was: a high net profit. To steer on this, data on greenhouse climate, irrigation, and crop are needed. For the crop, the most important data to steer on is the rate of leaf splitting."
What can one do with this insight?
"With every new leaf, a new cucumber grows. So, the faster the leaf splitting, the more cucumbers, which in turn generate income. Therefore, more information on leaf splitting allows for a good harvest forecast. However, leaf splitting is not yet automatically measured in practice. Usually, an employee goes into the greenhouse once a week to count the new leaves. While other data are continuously measured by sensors. That's why we developed our own method where leaf splitting can be continuously measured through image recognition."
The next question is: does cultivating based on data yield good results?
"We also investigated this. We cultivated cucumbers in three different scenarios: based on a cultivation plan, with autonomous control by AI, and by a Digital twin. The great thing is that the results were comparable. The AI model we used could not perform all tasks yet. For example, indicating how many cucumbers need to be thinned out was not part of the model. But the AI model could independently control the climate. Interestingly, cultivation experts found the temperature progression in that cultivation extreme, but the crop responded well to it. Autonomous growing is thus getting closer."
What now?
"The sector is already very advanced. But there are still challenges. In AGROS 2, which will start soon, we'll tackle some of these. Like: working with variable prices for products and energy, so AI can also steer on that. And increasing the reliability of the data. Because you have to be sure that what you're measuring is indeed correct to be able to regulate reliably on it"
So, will the entire sector switch to autonomous growing?
"That's not certain yet. But the most important thing is that a movement has been initiated in recent years. You can compare it to the closed greenhouse. Years ago, it was thought that all greenhouses would become closed. That didn't happen. But what did happen is the advent of The New Cultivation. That is a direct result of the research on the closed greenhouse back then. In this way, you can also view the development around autonomous growing. In recent years, for example, due to the discussions about autonomous growing, we've started to look differently at collecting and using data."
Source: Greenport West-Holland