Misiroot: A Robotic Minimum Invasion in Situ Imaging System for Plant Root Phenotyping

Zhihang Song, Purdue University

Abstract

Plant root phenotyping technologies play an important role in breeding, plant protection, and other plant science research projects. The root phenotyping customers urgently need technologies that are low-cost, in situ, non-destructive to the roots, and suitable for the natural soil environment. Many recently developed root phenotyping methods such as minirhizotron, CT, and MRI scanners have their unique advantages in observing plant roots, but they also have disadvantages and cannot meet all the critical requirements simultaneously. The study in this paper focuses on the development of a new plant root phenotyping robot that is minimally invasive to plants and working in situ inside natural soil, called “MISIRoot”. The MISIRoot system (patent pending) mainly consists of an industriallevel robotic arm, a mini-size camera with lighting set, a plant pot holding platform, and the image processing software for root recognition and feature extraction. MISIRoot can take high-resolution color images of the roots in soil with minimal disturbance to the root and reconstruct the plant roots’ three-dimensional (3D) structure at an accuracy of 0.1 mm. In a test assay, well-watered and drought-stressed groups of corn plants were measured by MISIRoot at V3, V4, and V5 stages. The system successfully acquired the RGB color images of the roots and extracted the 3D points cloud data which showed the locations of the detected roots in the soil. The plants measured by MISIRoot and plants not measured (controls) were carefully compared with Purdue’s Lilly 13-4 Hyperspectral Imaging Facility (reference). No significant differences were found between the two groups of plants at different growth stages. Therefore, it was concluded that MISIRoot measurements had no significant disturbance to the corn plant’s growth.

Degree

M.Sc.

Advisors

Jin, Purdue University.

Subject Area

Robotics|Design|Applied Mathematics|Electrical engineering|Industrial engineering|Mathematics|Medical imaging|Optics|Plant sciences

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS