WUR is working on Digital Twins for tomatoes, food and farming

News

WUR is working on Digital Twins for tomatoes, food and farming

Published on
January 22, 2020

Wageningen researchers are going to work on Digital Twins in the areas of tomatoes, food and farming. The projects are called Virtual tomato crops; Me, my diet and I; and Digital Future Farm. Digital Twins, one of the three investment themes of the WUR Strategic Plan 2019-2022, is a relatively new concept: computer models of individual objects or processes that are updated on the basis of real-time information.

WUR has searched for Digital Twin projects that can achieve major scientific and social breakthroughs in the Wageningen domains. Collaboration is a high priority for this investment programme, also because knowledge from different disciplines is essential for a successful Digital Twin.

Three subjects were chosen, on which groups of researchers from very diverse disciplines will work in the coming three years. Around 1.2m euros per project has been made available by WUR.

Read more about the three projects below:

Virtual tomato crops

Jochem Evers and his team are developing a digital twin of a tomato crop in a greenhouse: a 3D simulation model that is fed in real-time with sensor information from a real greenhouse. These constant updates make this digital twin more advanced than the existing simulation models. The interactions between the characteristics of the crop (the variety), the environmental factors and crop management are all simulated in the virtual crop. Because the model is linked to a real tomato crop in a greenhouse, it becomes possible to refine predictions more and more and thus make better choices for the real crop.

Read more about Virtual tomato crops

According to priority area leader Jochem Evers, “We chose the tomato because we already know so much about this crop; we have a lot of experience with measuring and 3D modelling tomatoes, as well as with various cultivation techniques. When developing the digital twin, we incorporate the ideas of a user group of interested people from the start of the project. They know what is going on in tomato cultivation and what we should focus on in the model”.

Evers hopes to have a properly working prototype in three years. Growers will be able to use it as a decision-support tool for growing tomatoes. For example, it will allow them to predict the effect of a cultivation measure on harvest and financial yield and make a decision for the real crop based on that prediction. The tool can also be used to determine strategies: one can make adjustments to the greenhouse settings in the model, and then carry out the most promising strategies on the real crop.

Growers will be able to use it as a decision-support tool for growing tomatoes. For example, it will allow them to predict the effect of a cultivation measure on harvest and financial yield and make a decision for the real crop based on that prediction

Evers talks about the benefits for research: “Researchers no longer have to do repeated tests to be able to make reliable statements, and you immediately know what the result of changes in, for example, greenhouse settings or pruning or varieties will be. That way we can also conduct targeted research into new varieties, finding out, for example, which qualities a variety should have to do well given any restrictions in energy consumption or in a certain type of greenhouse”.

Evers himself is a theoretical plant biologist, but his team is very diverse; employees of the (new) Netherlands Plant Eco-phenotyping Center (NPEC) on the campus together with WUR researchers in the fields of greenhouse horticulture, economics, agrotechnology, statistics and breeding. "That collaboration was already interesting in the start-up phase and the project hasn’t even begun."


Me, my diet and I

Lydia Afman and her multidisciplinary team are building a personalised digital twin to predict the rise in blood sugar (glucose), but especially the rise in blood fat (triglyceride) in the blood after a meal. Both are indicators of the risk of cardiovascular disease. There are no sensors for monitoring blood fat like there are for glucose – WUR has already done a lot of research into the latter. Based on already collected blood fat data from 500 overweight middle-aged people, the team will work together with researchers with knowledge of Artificial Intelligence (AI) to create a digital model. The predictive capacity of this digital twin will be tested in a human study and then improved.

Read more about Me, my diet and I

According to priority area leader Lydia Afman, associate professor of Human Nutrition and Health: “A person is also a system, just like a greenhouse, but every person responds differently in terms of blood fat. That is why we want to develop an algorithm based on this digital twin, with which we can predict the fat response and provide personal nutritional advice based on personal data, such as BMI, age, body fat distribution and, for example, blood pressure. We’re basically aiming for a personalised digital twin”.

Ultimately, the group wants to develop an app together with social scientists and dieticians, which provides nutritional advice in the form of recommended products or daily menu suggestions. The behaviour, preferences and values of the user must also be taken into account.

We want to develop an algorithm based on this digital twin, with which we can predict the fat response and provide personal nutritional advice based on personal data. We’re basically aiming for a personalised digital twin
Lydia Afman, associate professor Human Nutrition and Health

“It makes a lot of difference whether someone is a vegetarian, or is not allowed to have certain things because of his or her faith. Of course we want to develop something that works,” says Lydia.

According to her, the multidisciplinary collaboration of WUR employees is the strength of this project. “Our team includes nutritionists, a system / synthetic biologist, a toxicologist, economists with knowledge of the market and consumer behaviour, and researchers in food technology and bioinformatics. Everyone looks at the problem from their own discipline, and most are so specialised that you need each other to achieve a good result. The idea of a digital twin comes from the Strategic Plan, but is creatively implemented by the enthusiastic researchers. It’s very inspiring to be part of such an initiative”.


Digital Future Farm

Thomas Been and his team are working on a digital twin that displays the existing nitrogen cycle on an arable or dairy farm and with which researchers and farmers can determine how to close the nitrogen cycle as much as possible. A little less abstract: the required models are linked and provided with company data, sensor data, weather data and other observations of the company.

Read more about Digital Future Farm

Because data is different for all companies, the digital twin becomes specific for each company. The twin makes clear where it is necessary to supplement nitrogen or to prevent losses. Researchers can do scenario studies with this model, without having to test all scenarios in practice. Farmers know more quickly where they stand and what measures they should take that best suit their business. Ultimately, this project should lead to a kind of nitrogen dashboard, with which the researcher will be able to control the parameters and the farmer will be able to optimise the use of nitrogen at the farm.

If we develop a digital twin for nitrogen, we’ll also have an infrastructure that should be able to work with organic matter, water, phosphate and diseases and pests
Thomas Been, researcher

Priority area leaders Thomas Been and Claudia Kamphuis say that nitrogen was chosen because this substance is currently particularly interesting to growers, farmers and the government. “We also have many models that zoom in on nitrogen, such as growth models (when does the plant use how much nitrogen?) and soil models (with nitrogen binding and leaching depending on soil, crop, weather) and business management systems that supply the required nitrogen input, for example the nitrogen that has been brought onto the land as natural manure or fertiliser. We make these individual models ‘talk’ to each other to monitor nitrogen in the cycle, in order to limit losses and to minimise use”.

According to Been and Kamphuis, the possibilities are endless. “If we develop a digital twin for nitrogen, we’ll also have an infrastructure that should be able to work with organic matter, water, phosphate and diseases and pests”.

The development of the digital twin requires many different types of expertise. That is why plant, animal, soil and environmental scientists, economists and data scientists are working together on the development of the digital twin for the ‘digital farm of the future’.