Field-Based High-Throughput Phenotyping for Maize Plant Using 3D LiDAR Point Cloud Generated With a "Phenomobile".

Area of Science:

• Agricultural Engineering
• Robotics
• Plant Science

Background:

• Increasing global population necessitates enhanced crop production through advanced breeding techniques.
• Current crop phenotyping methods lack the speed and accuracy required for large-scale genetic improvement.
• High-throughput phenotyping is crucial for efficient crop breeding and development.

Purpose of the Study:

• To introduce a new field-based sensing solution for high-throughput crop phenotyping.
• To develop and validate a mobile robot system for non-invasive, rapid data collection in agricultural fields.
• To accurately measure key morphological parameters of maize plants using 3D laser scanning.

Main Methods:

• A mobile robot, termed 'phenomobile', equipped with a LiDAR sensor (Velodyne HDL64-S3) was utilized.
• Software developed using Robotic Operating System (ROS) facilitated data collection and analysis.
• Point cloud merging (using landmarks and Iterative Closest Points) and depth-band histograms with horizontal point density were employed for parameter extraction.

Main Results:

• The phenomobile system successfully collected 3D and 360° data for large plant groups efficiently.
• Accurate measurements of maize plant row spacing and individual plant height were achieved.
• The system demonstrated feasibility and efficiency in field-based high-throughput phenotyping experiments.

Conclusions:

• The proposed LiDAR-equipped mobile robot system provides a feasible and effective solution for high-throughput phenotyping.
• This approach overcomes the limitations of traditional phenotyping methods by enabling rapid, accurate, and non-invasive data acquisition.
• The technology supports advancements in crop breeding programs by providing essential morphological data at scale.