Improving Intelligent Crop Breeding with 3D Printed Sugar Beet Plant

A team of German researchers are working to bring farming into the future by developing AI-assisted crop pipeline improvement. By using laser scanning and consumer-grade FDM 3D printing, they were able to generate detailed 3D models of a sugar beet plant, which will help farmers support intelligent crop breeding—the science of improving agricultural plants to produce desired traits, such as being more nutritious, productive, and sustainable. The researchers, from University of Bonn and the Institute of Sugar Beet Research (IFZ), published their work in the journal GigaScience.

“This study addresses the importance of precise referencing in 3-dimensional (3D) plant phenotyping, which is crucial for advancing plant breeding and improving crop production. Traditionally, reference data in plant phenotyping rely on invasive methods. Recent advancements in 3D sensing technologies offer the possibility to collect parameters that cannot be referenced by manual measurements. This work focuses on evaluating a 3D printed sugar beet plant model as a referencing tool,” the team wrote.

Display of leaf models fitted to the point cloud of the reference model for (A) leaf 7 and (B) leaf 1. The marked lines indicate the sections used to measure leaf length, blade length, and blade width. More details about the algorithm are provided by Marks et al. (11).

In the face of ever more challenging conditions caused by environmental factors, it’s becoming increasingly important for farmers to adapt production of their crops in order to increase the level of production and ensure its security. Plant breeding has been known to help maintain stable yields and handle climate changes, but doing so for novel crop varieties is labor-intensive, time-consuming, and requires reliable information about the interaction of a plant’s genetic makeup, or genotype, with its environment. As the researchers explain, “The process of accessing this information in form of geometric or physiological properties of the plant is called plant phenotyping.”

Traditionally, humans would manually record information about the plants during phenotyping, such as the size of the crop, the shape and size of its leaves, fruit quality, and more. But these days, more phenotyping pipelines are using computer-assisted sensors, and often artificial intelligence, to automate the practice and more efficiently capture complex information about the plants. In fact, everything about planting crops is becoming more modern and automated, and the researchers said today’s plant breeding is a “data-centric enterprise, involving machine learning algorithms and sophisticated imaging technology to select desirable traits.”

“Advances in passive and active optical sensors, coupled with 3D reconstruction algorithms, provide the basis for precise high-throughput and high-resolution 3D analysis of above-ground plant structures, advancing 3D shoot phenotyping research [6, 7],” they wrote in their paper.

The proposed 3D printed reference model for sugar beet in (A) greenhouse and (B) field experiments.

Their goal in this research project was to create a better 3D model of the sugar beet plant that illustrates the main characteristics of the plant’s above-ground parts. This can be used as precise reference material to get more genetic information for guiding intelligent crop breeding, and can also assist with AI-driven crop improvement pipelines.

“In the field of three-dimensional plant phenotyping, the referencing of utilized sensor systems, computer algorithms and captured morphological parameters represents a challenging yet fundamentally important task,” explained Jonas Bömer, a PhD student at IFZ. “The application of additive manufacturing technologies for the generation of reproducible reference models presents a novel opportunity to develop standardized methodologies for objective and precise referencing, thereby benefiting both scientific research and practical plant breeding.”

For a “standard plant,” sensors need data on all relevant characteristics, even complex 3D traits such as leaf orientation or “the shadow cast of a plant.” Just like in surgical planning or medical training, a life-sized, three-dimensional model can be much more helpful than computer data and two-dimensional representations. The research team says that the 3D printable sugar beet plant models they developed can be easily reproduced for use in the field, which in this case is an actual field, or a greenhouse.

Workflow for the creation of the artificial sugar beet reference plant. The intermediate processes are shown in white boxes. Conducted production steps are highlighted in orange while the used software and hardware are pictured in light red.

The team used laser-based light detection and ranging (LiDAR) technology to gather data for their model, and scanned a sugar beet plant with a Faro Focus system to generate 3D data from 12 angles. They processed all the data using Faro Scene software, the Open3D Python library, open source CloudCompare, Blender, and PrusaSlicer. Then, they fabricated a life-sized model of the plant out of PETG using a Prusa i3 MK3S+ 3D printer.

“Supports were activated on the build plate, and the 0.15-mm quality profile was chosen and adjusted to our requirements (perimeters: 3, fill density: 50%, fill pattern: Gyroid, brim width: 5 mm, overhang threshold: 30°),” the researchers wrote.

The model was then tested as a reference point, first in the laboratory and then in the field. Because these models can be easily and accurately produced using 3D printing, the team says that they “enable standardized reference approaches in 3D plant phenotyping.” To ensure this, they made the files available for download and reuse via Thingiverse, as well as three collections in Sketchfab. This way, other scientists can create their own copies of the sugar beet reference plant and make research from multiple labs more comparable.

The research team hopes that their study shows how modern technologies like 3D printing can help with breeding for other types of cultivated crops, too.

“The value in a printable 3D model is that you can print multiple copies, one per field of crops,” explained GigaScience Data Scientist Chris Armit. “As a low-cost phenotyping strategy, where the major cost is the LIDAR scanner, it would be fantastic to see this approach tested on other crops such as rice or African orphan crops, where there is a need for low-cost phenotyping solutions.”

“The proposed reference model is accurately reproducible and stable over a longer period, indicating that FDM 3D printing is a suitable production technique for the suggested applications. The introduced process of creating a 3D reference model for sugar beets can serve as an example for developing similar reference models for other widely used arable or horticultural crops,” the researchers concluded.