Development and Performance Evaluation of a Plant Phenotyping Platform Using Low-cost IoT Devices

Abstract

Advancement of plant breeding and cultivation management technologies requires technology to quantitatively measure and evaluate plant growth and environmental responses. Recently, with the remarkable development of ICT (Information and Commu- nications Technology), research on the development of phenotyping technology that enables rapid measurement of plant growth characteristics has been conducted. In this study, we developed a plant phenotyping robot that can automatically measure plant growth information by using low-cost IoT (Internet of Things) devices as well as open-source image processing and analysis li- braries. The developed robot runs on rails placed on both sides of a hydroponics bed. It can automatically measure plant growth images according to a schedule by controlling the robot’s running and positioning with AR (Augmented Reality) markers. We evaluated the performance of the robot as a plant phenotyping robot by taking images of the growth of spinach from immediately after the planting of young seedlings to harvesting every 3 h with an RGB-D camera mounted on the robot. The position control performance of the developed robot was about 3 mm. The spatio-temporal RGB and depth information for the entire cultivation bed had high enough accuracy to estimate growth features and defects of leafy vegetables. Keywords