Development of a high-throughput field phenotyping rover optimized for size-limited breeding fields as open-source hardware

Abstract

Phenotyping is a critical process in plant breeding, especially when there is an increasing demand for streamlining a selection process in a breeding program. Since manual phenotyping has limited efficiency, high-throughput phenotyping methods are recently popularized owing to progress in sensor and image processing technologies. However, in a size-limited breeding field, which is common in Japan and other Asian countries, it is challenging to introduce large machinery in the field or fly unmanned aerial vehicles over the field. In this study, we developed a ground-based high-throughput field phenotyping rover that could be easily introduced to a field regardless of the scale and location of the field even without special facilities. We also made the field rover open-source hardware, making its system available to public for easy modification, so that anyone can build one for their own use at a low cost. The trial run of the field rover revealed that it allowed the collection of detailed remote-sensing images of plants and quantitative analyses based on the images. The results suggest that the field rover developed in this study could allow efficient phenotyping of plants especially in a small breeding field.

Keywords: field phenotyping rover; high-throughput phenotyping; image processing; open-source hardware; proximal sensing.

Fig. 1.

The drawing for our field phenotyping rover by the 3D CAD software SolidWorks (Dassault Systemes SE, Vélizy-Villacoublay, France). The 45-mm² aluminum frame was assembled and equipped with components for control and driving. The CAD data file is available our GitHub repository.

Fig. 2.

The hardware and software scheme.

Fig. 3.

Transmitter configurations.

Fig. 4.

Images captured by our field phenotyping rover on the field trial. (A–C) Images from center, right, and left camera, respectively. Six different cultivars each consisted of 10 individual plants are observed. Photo was taken on April 5, 2021. (D) An image from the top camera taken at the same location as (A–C), but in the heading season. Wheat spikes, as well as flowers (bottom) and awns (right) when magnified, are clearly visible. Photo was taken on April 30, 2021 and cropped from the original image for the purpose of visibility here.

Fig. 5.

Follow-up and retrospective observation of physiological disorder. All of the images are taken approximately at the same location over a time series. The disorder was noticed in late March, although it first appeared earlier that month. It eventually disappeared as new healthy leaves grew in April. White arrows point at an instance of disorder observable in the image.

Fig. 6.

Example of automatic plot extraction using EasyIDP. Red lines denote a border between adjacent plots.

Fig. 7.

Quantitative analysis on the progression of heading using deep learning models. (A) Number of heads detected by YOLOv5. (B) Comparison of heading dates between field manual observation and image analysis.

Fig. 8.

Test on slope climbing capability. The ridge between fields were roughly 30 cm higher than the field surface level. The slope was approximately 2 m long, making the slope gradient 15%. The slopes were made of aluminum and easily carried by one person.