Unmanned Aerial Vehicles for High-Throughput Phenotyping and Agronomic Research

Abstract

Advances in automation and data science have led agriculturists to seek real-time, high-quality, high-volume crop data to accelerate crop improvement through breeding and to optimize agronomic practices. Breeders have recently gained massive data-collection capability in genome sequencing of plants. Faster phenotypic trait data collection and analysis relative to genetic data leads to faster and better selections in crop improvement. Furthermore, faster and higher-resolution crop data collection leads to greater capability for scientists and growers to improve precision-agriculture practices on increasingly larger farms; e.g., site-specific application of water and nutrients. Unmanned aerial vehicles (UAVs) have recently gained traction as agricultural data collection systems. Using UAVs for agricultural remote sensing is an innovative technology that differs from traditional remote sensing in more ways than strictly higher-resolution images; it provides many new and unique possibilities, as well as new and unique challenges. Herein we report on processes and lessons learned from year 1-the summer 2015 and winter 2016 growing seasons-of a large multidisciplinary project evaluating UAV images across a range of breeding and agronomic research trials on a large research farm. Included are team and project planning, UAV and sensor selection and integration, and data collection and analysis workflow. The study involved many crops and both breeding plots and agronomic fields. The project's goal was to develop methods for UAVs to collect high-quality, high-volume crop data with fast turnaround time to field scientists. The project included five teams: Administration, Flight Operations, Sensors, Data Management, and Field Research. Four case studies involving multiple crops in breeding and agronomic applications add practical descriptive detail. Lessons learned include critical information on sensors, air vehicles, and configuration parameters for both. As the first and most comprehensive project of its kind to date, these lessons are particularly salient to researchers embarking on agricultural research with UAVs.

Fig 1. Gartner hype cycle cartoon of the subjective value and development stage of various technologies discussed here.

Most of the base technologies are mature and productive, but the integration of all of these technologies is new and likely over-hyped.

Fig 2. Generalization of the integration of teams and responsibilities in the TAMU-UAS project.

Field researchers are end users and primarily involved in experimental design and ground-truthing data. Aerospace and mechanical engineers, and ecosystem scientists are primarily involved in raw UAS data collection. Geospatial scientists serve as a clearinghouse for the UAS data and also perform mosaicking—the stitching together of many small images to build one ortho-rectified and radiometrically seamless large image. Agricultural engineers are the nexus of the project, turning the UAS data into actionable results for the end users. The administration team provides and manages funds, facilitates meetings, and coordinates communications and initiatives.

Fig 3. Workflow to start a UAV project for phenotyping and agronomic research, from interdisciplinary team establishment to decision making.

Fig 4. Experiment site—Brazos Bottom research farm.

It was divided into route-packs (eight larger polygons in yellow) to be covered efficiently across different PI’s fields during an individual flight by fixed-wing UAVs. An individual flight of the multi-rotor UAV was based on a single field from a single researcher marked by smaller black polygons in the figure.

Fig 5. External physical characteristics of UAVs used in this study.

(A) RMRC Anaconda fixed-wing vehicle. (B) PrecisionHawk Lancaster fixed-wing vehicle. (C) TurboAce X88 octocopter with autopilot computer interface (left), multiple battery options (middle), and navigation and gimbal transmitters (right). The individuals appearing in these figures gave written informed consent (as outlined in PLOS consent form) to publish these pictures.

Fig 6. Sensors carried by the UAVs used in this study.

(A) Sentek GEMS multispectral camera carried by the Anaconda fixed-wing UAV. (B) Nikon J3 digital camera (left) and modified multispectral camera (right) carried by the Lancaster fixed-wing UAV. (C) DJI P3-005 4K camera carried by the X88 octocopter.

Fig 7. Relationship between flight and sensor parameters used to determine optimal flight and sensor configurations before flights (use Anaconda fixed-wing and Sentek multispectral camera as an example here).

(A) GSD and flying altitude AGL under a fixed sensor FOV (26.31° vertically). (B) Image overlap and UAV ground speed under a fixed flying altitude (120 m) and a fixed sensor frame rate (1.4 s/frame).

Fig 8. Three different flight paths over a 30-ha route pack evaluated by Anaconda fixed-wing UAV in this study.

(A) Standard parallel flight path. (B) Cross-stitch flight path. (C) Moving-box flight path. The yellow lines represent planned flight paths; the green balloon-shape icons represent waypoints along the flight path for GPS navigation.

Fig 9. Two types of GCPs were used in this study.

(A) A set of semi-permanent painted concrete tiles with 10%, 20% and 40% reflectance for radiometric correction. (B) A set of semi-permanent GCPs as seen in the NIR and RGB images collected with a fixed-wing UAV at 120 m AGL. (C) A portable wooden frame GCP covered with canvas painted with a double-cross pattern. (D) A portable GCP as imaged with the octocopter at 15 m AGL.

Fig 10. Results of plant height estimates.

Digital surface models (DSMs) and correlations between aerial estimated plant height and ground truth plant height on maize (A) based on 705 observations (C), and on sorghum (B) based on 40 observations (D).

Fig 11. NDVI map generated from multispectral data collected with the Sentek sensor onboard the Anaconda fixed wing UAV platform.

Fig 12. Results of winter wheat biophysical study.

(A) Correlation between wheat leaf area index (LAI) measured on the ground using leaf area meter and NDVI calculated from aerial imagery. (B) Correlation between wheat ground cover estimated on the ground and NDVI calculated from aerial imagery.

Fig 13. Results of soil and plant interaction study.

(A) Normalized difference vegetation index (NDVI) map thresholded to remove bare soil. (B) Apparent electrical conductivity (EC_a) map of the soil. (C) Correlation between NDVI and seed cotton yield at late growth stage. (D) Correlation between thresholded NDVI and seed cotton yield.

Fig 14. Correlation between thresholded NDVI and soil apparent electrical conductivity (EC_a).

Fig 15. Results of the weed management evaluation study.

(A) Aerial image mosaic of a weed management experiment with 28 plots. (B) Classification of soil (brown) and vegetation (green). (C) Comparison between estimated weed control from aerial imagery and ground truth weed control of each treatment. (D) Correlation between estimated weed control from aerial imagery and ground truth weed control based on 28 observations.