Design of a Robotic Platform for Plant Root Imaging and Biomass Estimation
Author(s)
Kliman, Randall J.
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
Plant health and analysis has historically been a difficult undertaking due to both cost barriers and complex experimental setup. Our project focuses on analyzing the growth of lettuce plants grown in hydroponic systems. One of the main indicators of plant health is root biomass which is difficult to measure without the removal of the plant from its growth medium and destruction of the roots and leaves. In this work, we discuss the creation of an automated system for the estimation of plant root and eventually shoot biomass over its lifecycle. By using a robotics and computer vision-based system, we are able to estimate the biomass of a plant nondestructively. This system uses a robotic arm and turntable to capture 40 images at equidistant angles around a hydroponically grown plant. These images are then processed into silhouettes and used in a voxel-based volumetric 3D reconstruction method which outputs the resulting 3D model. We use a space carving method with a raytracing based optical correction method to create high accuracy 3D models. Analysis of this model shows that the underlying algorithm reconstructs the plant root structure accurately and gives a robust measurement of the plant root volume, which can be calibrated to indicate biomass.
Sponsor
Date
2022-12-15
Extent
Resource Type
Text
Resource Subtype
Thesis