Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Simulation of an In-field Phenotyping Robot: System Design, Vision-based Navigation and Field Mapping

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan

Citation:  2022 ASABE Annual International Meeting  2200999.(doi:10.13031/aim.202200999)
Authors:   Zhengkun Li, Rui Xu, Changying Li, Longsheng Fu
Keywords:   Phenotyping robot; Navigation; Multi-object tacking; ROS; Gazebo; Simulation; Agriculture

Abstract. Cultivating high-yield and high-quality crops is important for addressing the growing demand in food and fiber from a growing population. In selective breeding programs, autonomous robotic systems have been gradually replacing the manual phenotypic trait measurements which are time-consuming and labor-intensive with relatively low temporal and spatial resolutions. In this paper, we present a Robot Operating System (ROS)-based phenotyping robot “MARS-PhenoBot” and demonstrate its reliable capacities including vision-based navigation and crop mapping through ROS-Gazebo simulation. MARS-PhenoBot is a solar-powered modular platform with a four-wheel steering and four-wheel driving configuration. Each wheel module is equipped with an independent suspension mechanism, making it adaptable to uneven field terrain. We developed a navigation strategy that relies on the crop row features extracted from images and exploits the prior knowledge of crop arrangement to autonomously navigate in cotton fields without any explicit map. For field mapping, a 2D map containing the distribution of weeds and cotton plants was developed based on a multi-object tracking algorithm and 2D-to-3D projection transform. The tracking algorithm that combines YOLO detector and DeepSORT tracker was applied to identify each plant in the field. In addition, a 3D mapping approach based on RTAB-Map SLAM method is deployed in our mapping architecture to generate the 3D map of cotton field. The simulation results show that our proposed navigation and crop mapping algorithms are capable of automated robotic phenotyping for field crops. The methodology developed in this study is scalable and can be easily deployed to the real agricultural robots to perform automated phenotyping tasks for crop field.

(Download PDF)    (Export to EndNotes)