3D Robotic System Development for High-throughput Crop Phenotyping

摘要

Abstract: Plant breeding programs are working towards developing new high-yielding crop varieties to accommodate the increasing demand for food. However, high-throughput phenotyping remains to be the bottleneck that is currently limiting the complete breeding potential. In this project, a 3D robotic system was developed to conduct automated high-throughput phenotyping in cereal crops. The 3D robotic phenotyping system consisted of an aluminum framework to support a 3D sliding system (sliders and tracks), which allows a sensor mount travel in X and Y axis in a selected height (Z axis). The system was controlled with a custom designed algorithm based on LabVIEW program. A control box was used to interface the system with a computer. During preliminary evaluation, a thermal camera and a multispectral camera were installed on the sensor mount, and the integrated automated phenotyping system was continuously operated for 48 hours for autonomous data collection. The 3D robotic system had been working precisely based on the design specifications. Results showed that the 3D robotic system had time repeatability with trigger activation within 4 s and positioning error less than 0.78 mm, indicating the potential of system for automated, systematic high-throughput phenotyping.