Automated in-field leaf-level hyperspectral imaging of corn plants using a Cartesian robotic platform

Abstract

Hyperspectral Imaging (HSI) has been widely adopted in field plant phenotyping activities. Current HSI solutions such as airborne remote sensing platforms and handheld spectrometers have been proven effective and have become popular in various phenotyping applications. However, the imaging quality of current airborne sensing systems still suffers from various noises due to the changing ambient lighting condition, long imaging distance, and comparatively low resolution. Handheld leaf spectrometers provide a higher quality of spectral data, but they only measure a small spot on the leaf, which cannot represent the whole leaf or canopy very well due to the great variation between different locations. In 2018, the Purdue Ag engineers developed a new handheld hyperspectral leaf imager, LeafSpec. For the first time, phenotyping researchers were able to collect high-resolution hyperspectral leaf images without the impacts of the changing ambient light and leaf slopes. However, the application of LeafSpec was still limited by its low throughput and intensive labor cost in the field measurements. The goal of this project was to develop a robotic system that could replace the human operator to perform in-field and leaf-level HSI using LeafSpec. The system consisted of a machine-operable version of the LeafSpec device, a machine vision system for target leaf detection, and a customized cartesian robotic manipulator with five Degrees of Freedom (DOF). In the 2019 field test, the designed system collected data from corn plants with two genotypes and three levels of nitrogen treatments with an average cycle time of 86 s. The nitrogen content predicted by the designed system had an R² value of 0.7307 against the ground truth. The prediction could also differentiate the different nitrogen treatments with P-values of 0.0193 and 0.0102. The performance was similar to human operators’. The developers, therefore, conclude that the robotic system has the potential of replacing human operators for LeafSpec hyperspectral corn leaf imaging in the field.

Plant phenotyping relevance

トウモロコシ葉のハイパースペクトル表現型取得を自動化するロボットシステムを開発し、窒素含量推定性能を検証しており、表現型取得・抽出法が中心である。

Code and data availability

公開論文であることは確認できましたが、現在の公式API・許可済み取得経路では本文を自動取得できませんでした。

No evidence-backed public reproduction asset is currently recorded.