Digital twin project ‘Virtual Tomato Crops’
The digital twin can be used to predict growth and development of tomato plants in response to real-time environmental factors and management decisions. This allows for better informed decisions regarding the agronomic management. Also, the digital twin can be used to inform tomato breeding by identifying key target traits. In the digital twin, each tomato plant in the crop is represented in 3D, integrating a set of traits that correspond to model parameters. Thereby, the twin enables users to predict crop responses (growth, development and production) to greenhouse and management conditions that affect production efficiency, light intensity and quality, CO2 dosing, nutrient availability and leaf pruning.
The output of the model, which is updated as the real crop grows and develops, can be used for automatic control of greenhouse climate settings, following a model predictive control strategy. Research questions to be answered concern effects of model granularity on climate control advice, and the effect of daily crop status update on control performance in terms of light use efficiency. Furthermore, the digital twin can be used to virtually explore leaf pruning strategies, to test different greenhouse cover types, and to select superior crop traits.
The ultimate goal of this digital twin is to increase resource use efficiency of greenhouse tomato systems, resulting in lower dependence on external energy inputs, a further reduction in CO2 emissions and optimization of water use and fertigation. This will reduce costs and reliance on inputs, making tomato growing more economical. The digital twin will work towards production of greenhouse tomatoes with a minimum of resources as well as demand driven by consumer preferences, and thus will realise a feasible cultivation and production system. The VTC project is the first step towards this ambitious goal.
VTC aims for a feature real-time updating of plant parameters and environmental variables, based on high-tech sensor equipment available in the NPEC greenhouse facilities. During two experiments inside a compartment with conveyor belts, data was collected using the MaxiMarvin for 3D plant reconstruction, RGB side view camera, thermal camera. The climate sensors measured the desired quantities directly. The image raw sensor data were processed to estimate plant traits using deep-learning methods.
Stakeholder communication throughout the project is done by Marc-Jeroen Bogaardt (Wageningen Economic Research). Contact person for development of phenotyping protocols and estimation of model parameters is Gert Kootstra (Farm Technology). Jochem Evers (Centre for Crop Systems Analysis) is the contact person for development and integration of the tomato plant model, and Simon van Mourik (Farm Technology) leads efforts on model application on short- and longer-term decision support. Communication on the use of the NPEC facilities within the project goes through Rick van de Zedde (Greenhouse Technology).
Essential further expertise is contributed by:
Daniela Bustos Korts at Biometris
Pieter de Visser at Greenhouse Horticulture
Tim van Daalen and Gert-Jan Swinkels at Greenhouse Technology
Elias Kaiser and Nastassia Vilfan at Horticulture and Product Physiology
Peter Roos at the Laboratory of Genetics
Harm Bartholomeus at the Laboratory of Geo-information Science and Remote Sensing
Patrick Hendrickx at Plant Breeding