An Autonomous Ground System for 3D LiDAR-based Crop Scouting

Abstract

<b>Abstract.</b> Remote sensing techniques provide researchers with efficient tools to monitor field evolution and crop development through the season. However, to identify plant- or genotype-specific variability across the field it is usually still necessary to inspect the crop from the ground. Ground truthing provides reliable measurements with which to compare any cause of variability found in the data, but it requires manual, time-consuming, and usually destructive sampling tasks. Additional work on finding new approaches to reduce the burden and allow researchers and plant scientists to devote their time to a more meaningful work is still necessary. In this study we present a robotic system for autonomous in-field 3D LiDAR data on plot-based breeding trials. The robotic platform consisted of a Husky A200 robotic platform equipped with a FARO Focus S70 3D laser scanner and was able to perform autonomous field measurements of relevant plant features without human intervention. The robotic system was integrated under the ROS framework to centralize the control of operation and to ensure the interoperability of all the components. The system provided realistic three-dimensional data extracted directly from the crop, which will allow the characterization of plant traits at the plot level and plant level—plant architecture, height and volume, and cotton boll mapping—accurately. This work demonstrates an autonomous phenotyping platform that leverages two off-the-shelf equipment for the quantitative assessment of in-field cotton plant traits. This work can facilitate the analysis of both large agricultural fields and small breeding trials to contribute to the advancement of plant phenomics.