Crop improvement programs require large and meticulous selection processes that effectively and accurately collect and analyze data to generate quality plant products as efficiently as possible, develop superior cropping and/or crop improvement methods. Typically, data collection for such testing is performed by field teams using hand-held instruments or manually-controlled devices. Although steps are taken to reduce error, the data collected in such manner can be unreliable due to human error and fatigue, which reduces the ability to make accurate selection decisions. Monsanto engineering teams have developed a high-clearance mobile platform (Rover) as a step towards high throughput and high accuracy phenotyping at an industrial scale. The rovers are equipped with GPS navigation, multiple cameras and sensors and on-board computers to acquire data and compute plant vigor metrics per plot. The supporting IT systems enable automatic path planning, plot identification, image and point cloud data QA/QC and near real-time analysis where results are streamed to enterprise databases for additional statistical analysis and product advancement decisions. Since the rover program was launched in North America in 2013, the number of research plots we can analyze in a growing season has expanded dramatically. This work describes some of the successes and challenges in scaling up of the rover platform for automated phenotyping to enable science at scale.
Near-infrared (NIR) spectroscopic measurement of blood and tissue chemistry often requires a large set of subject data
for training a prediction model. We have previously developed the principal component analysis loading correction
(PCALC) method to correct for subject related spectral variations. In this study we tested the concept of developing
PCALC factors from simulated spectra. Thirty, two-layer solid phantoms were made with 5 ink concentrations (0.004%-
0.02%), 2 μs' levels, and 3 fat thicknesses. Spectra were collected in reflectance mode and converted to absorbance by
referencing to a 99% reflectance standard. Spectra (5733) were simulated using Kienle's two-layer turbid media model
encompassing the range of parameters used in the phantoms. PCALC factors were generated from the simulated spectra
at one ink concentration. Simulated spectra were corrected with the PCALC factors and a PLS model was developed to
predict ink concentration from spectra. The best-matched simulated spectrum was identified for each measured phantom
spectrum. These best-matched simulated spectra were corrected with the PCALC factors derived from the simulated
spectra set, and they were used in the PLS model to predict ink concentrations. The ink concentrations were predicted
with an R2=0.897, and an estimated error (RMSEP) of 0.0037%. This study demonstrated the feasibility of using
simulated spectra to correct for inter-subject spectral differences and accurately determine analyte concentrations in
turbid media.
Noninvasive near infrared (NIR) spectroscopic measurement of muscle oxygenation requires the penetration of
light through overlying skin and fat layers. We have previously demonstrated a dual-light source design and
orthogonalization algorithm that corrects for inference from skin absorption and fat scattering. To achieve
accurate muscle oxygen saturation (SmO2) measurement, one must select the appropriate source-detector
distance (SD) to completely penetrate the fat layer. Methods: Six healthy subjects were supine for 15min to
normalize tissue oxygenation across the body. NIR spectra were collected from the calf, shoulder, lower and
upper thigh muscles with long SD distances of 30mm, 35mm, 40mm and 45mm. Spectral preprocessing with the
short SD (3mm) spectrum preceded SmO2 calculation with a Taylor series expansion method. Three-way
ANOVA was used to compare SmO2 values over varying fat thickness, subjects and SD distances. Results:
Overlying fat layers varied in thickness from 4.9mm to 19.6mm across all subjects. SmO2 measured at the four
locations were comparable for each subject (p=0.133), regardless of fat thickness and SD distance. SmO2
(mean±std dev) measured at calf, shoulder, low and high thigh were 62±3%, 59±8%, 61±2%, 61±4%
respectively for SD distance of 30mm. In these subjects no significant influence of SD was observed (p=0.948).
Conclusions: The results indicate that for our sensor design a 30mm SD is sufficient to penetrate through a
19mm fat layer and that orthogonalization with short SD effectively removed spectral interference from fat to
result in a reproducible determination of SmO2.
Conference Committee Involvement (4)
Autonomous Air and Ground Sensing Systems for Agricultural Optimization and Phenotyping V
27 April 2020 | Online Only, California, United States
Autonomous Air and Ground Sensing Systems for Agricultural Optimization and Phenotyping IV
15 April 2019 | Baltimore, MD, United States
Autonomous Air and Ground Sensing Systems for Agricultural Optimization and Phenotyping III
18 April 2018 | Orlando, FL, United States
Autonomous Air and Ground Sensing Systems for Agricultural Optimization and Phenotyping II
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.