CherryTwin: an exemplary concept for digital twins in horticulture

Digitalization in horticulture and orchards is a challenging topic that becomes increasingly important in research. Exemplified by the use case of sweet cherry trees (Prunus avium ‘Satin’), we present the "For5G" project, where a framework for the creation of digital twins for individual trees is developed. Multiple twins can be aggregated to gain detailed insights into arbitrary sized groups of trees e.g., entire orchards. Starting with a concept for data collection using a photogrammetry drone equipped with a 5G module, we present an end-to-end pipeline for digital twins. Stored in a knowledge graph database, AI was utilized for various detailed evaluation pipelines solving current problems in this domain: this includes photorealistic 3D reconstructions, automated tree skeleton extraction and fruit localization and counting. After following three sweet cherry trees over the 2023 season, we published a data set and present an outlook of ongoing work like biomass estimation and early yield prediction. Additionally, under development is an app for end-users like farmers, researchers or breeders to visualize all aspects of the digital twins aggregated in arbitrary scale. This publication shifts the focus from detailed technical implementations of singular aspects to the big picture and combines previous project results with current plans to show the potential of digital twins in horticulture. Using real world data and realistic use-cases, an exemplary concept for digital twins in horticulture is transparently displayed in a way that can be adapted to comparable use cases or similar domains.