This PDF file contains the front matter associated with SPIE Proceedings Volume 10218, including the Title Page, Copyright information, Table of Contents, Introduction (if any), and Conference Committee listing.

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.

AGERpoint is defining a new technology space for the growers’ industry by introducing novel applications for sensor technology and data analysis to growers of permanent crops. Serving data to a state-of-the-art analytics engine from a cutting edge sensor platform, a new paradigm in precision agriculture is being developed that allows growers to understand the unique needs of each tree, bush or vine in their operation. Autonomous aerial and terrestrial vehicles equipped with multiple varieties of remote sensing technologies give AGERpoint the ability to measure key morphological and spectral features of permanent crops. This work demonstrates how such phenotypic measurements combined with machine learning algorithms can be used to determine the variety of crops (e.g., almond and pecan trees). This phenotypic and varietal information represents the first step in enabling growers with the ability to tailor their management practices to individual plants and maximize their economic productivity.

Our group has developed a custom, multi-modal sensor suite and data analysis pipeline to phenotype crops in the field using unpiloted aircraft systems (UAS). This approach to high-throughput field phenotyping is part of a research initiative intending to markedly accelerate the breeding process for refined energy sorghum varieties. To date, single rotor and multirotor helicopters, roughly 14 kg in total weight, are being employed to provide sensor coverage over multiple hectaresized fields in tens of minutes. The quick, autonomous operations allow for complete field coverage at consistent plant and lighting conditions, with low operating costs.

The sensor suite collects data simultaneously from six sensors and registers it for fusion and analysis. High resolution color imagery targets color and geometric phenotypes, along with lidar measurements. Long-wave infrared imagery targets temperature phenomena and plant stress. Hyperspectral visible and near-infrared imagery targets phenotypes such as biomass and chlorophyll content, as well as novel, predictive spectral signatures. Onboard spectrometers and careful laboratory and in-field calibration techniques aim to increase the physical validity of the sensor data throughout and across growing seasons. Off-line processing of data creates basic products such as image maps and digital elevation models. Derived data products include phenotype charts, statistics, and trends.

The outcome of this work is a set of commercially available phenotyping technologies, including sensor suites, a fully integrated phenotyping UAS, and data analysis software. Effort is also underway to transition these technologies to farm management users by way of streamlined, lower cost sensor packages and intuitive software interfaces.

In order to address the worldwide growing demand for food, agriculture is facing certain challenges and limitations. One of the important threats limiting crop productivity is salinity. Identifying salt tolerate varieties is crucial to mitigate the negative effects of this abiotic stress in agricultural production systems. Traditional measurement methods of this stress, such as biomass retention, are labor intensive, environmentally influenced, and often poorly correlated to salinity stress alone. In this study, hyperspectral imaging, as a non-destructive and rapid method, was utilized to expedite the process of identifying relatively the most salt tolerant line among four wheat lines including Triticum aestivum var. Kharchia, T. aestivum var. Chinese Spring, (Ae. columnaris) T. aestivum var. Chinese Spring, and (Ae. speltoides) T. aestivum var. Chinese Spring. To examine the possibility of early detection of a salt tolerant line, image acquisition was started one day after stress induction and continued on three, seven, and 12 days after adding salt. Simplex volume maximization (SiVM) method was deployed to detect superior wheat lines in response to salt stress. The results of analyzing images taken as soon as one day after salt induction revealed that Kharchia and (columnaris)Chinese Spring are the most tolerant wheat lines, while (speltoides) Chinese Spring was a moderately susceptible, and Chinese Spring was a relatively susceptible line to salt stress. These results were confirmed with the measuring biomass performed several weeks later.

This paper presents the work being conducted at Cal Poly Pomona on the collaboration between unmanned aerial and ground vehicles for precision agriculture. The unmanned aerial vehicles (UAVs), equipped with multispectral/hyperspectral cameras and RGB cameras, take images of the crops while flying autonomously. The images are post processed or can be processed onboard. The processed images are used in the detection of unhealthy plants. Aerial data can be used by the UAVs and unmanned ground vehicles (UGVs) for various purposes including care of crops, harvest estimation, etc. The images can also be useful for optimized harvesting by isolating low yielding plants. These vehicles can be operated autonomously with limited or no human intervention, thereby reducing cost and limiting human exposure to agricultural chemicals. The paper discuss the autonomous UAV and UGV platforms used for the research, sensor integration, and experimental testing. Methods for ground truthing the results obtained from the UAVs will be used. The paper will also discuss equipping the UGV with a robotic arm for removing the unhealthy plants and/or weeds.

The rapid advancement and availability of small unmanned aircraft systems (sUAS) has led to many novel exploitation tasks utilizing that utilize this unique aerial imagery data. Collection of this unique data requires novel flight planning to accomplish the task at hand. This work describes novel flight planning to better support structure-from-motion missions to minimize occlusions, autonomous and periodic overflight of reflectance calibration panels to permit more efficient and accurate data collection under varying illumination conditions, and the collection of imagery data to study optical properties such as the bidirectional reflectance distribution function without disturbing the target in sensitive or remote areas of interest. These novel mission planning algorithms will provide scientists with additional tools to meet their future data collection needs.

Thanks to the development of camera technologies, small unmanned aerial systems (sUAS), it is possible to collect aerial images of field with more flexible visit, higher resolution and much lower cost. Furthermore, the performance of objection detection based on deeply trained convolutional neural networks (CNNs) has been improved significantly. In this study, we applied these technologies in the melon production, where high-resolution aerial images were used to count melons in the field and predict the yield. CNN-based object detection framework-Faster R-CNN is applied in the melon classification. Our results showed that sUAS plus CNNs were able to detect melons accurately in the late harvest season.

Crop yield estimation is an important task in product management and marketing. Accurate yield prediction helps farmers to make better decision on cultivation practices, plant disease prevention, and the size of harvest labor force. The current practice of yield estimation based on the manual counting of fruits is very time consuming and expensive process and it is not practical for big fields. Robotic systems including Unmanned Aerial Vehicles (UAV) and Unmanned Ground Vehicles (UGV), provide an efficient, cost-effective, flexible, and scalable solution for product management and yield prediction. Recently huge data has been gathered from agricultural field, however efficient analysis of those data is still a challenging task. Computer vision approaches currently face diffident challenges in automatic counting of fruits or flowers including occlusion caused by leaves, branches or other fruits, variance in natural illumination, and scale. In this paper a novel deep convolutional network algorithm was developed to facilitate the accurate yield prediction and automatic counting of fruits and vegetables on the images. Our method is robust to occlusion, shadow, uneven illumination and scale. Experimental results in comparison to the state-of-the art show the effectiveness of our algorithm.

Even in the nascent stage of unmanned aerial vehicles, it was projected that UAV technologies would closely integrate into agriculture activities at a fast pace and become a ubiquitous and low-cost tool for farming operations. Nevertheless, several years later, it is widely recognized that available UAV technology has not yet integrated into agriculture as expected despite multiple UAV platform offerings. In this paper, we discuss several concepts related to the commercial integration of UAV technology for agriculture, along with expected or existing solutions that can further its development and acceptance, especially in the US. Examples from the AggieAir UAV Research Group, at Utah State University, are included to provide a better understanding of the presented concepts.

Though sharing with many commonalities, one of the major differences between conventional high-altitude airborne remote sensing and low-altitude unmanned aerial system (UAS) based remote sensing is that the latter one has much smaller ground footprint for each image shot. To cover the same area on the ground, it requires the low-altitude UASbased platform to take many highly-overlapped images to produce a good mosaic, instead of just one or a few image shots by the high-altitude aerial platform. Such an UAS flight usually takes 10 to 30 minutes or even longer to complete; environmental lighting change during this time span cannot be ignored especially when spectral variations of various parts of a field are of interests. In this case study, we compared the visible reflectance of two aerial imagery – one generated from mosaicked UAS images, the other generated from a single image taken by a manned aircraft – over the same agricultural field to quantitatively evaluate their spectral variations caused by the different data acquisition strategies. Specifically, we (1) developed our customized ground calibration points (GCPs) and an associated radiometric calibration method for UAS data processing based on camera’s sensitivity characteristics; (2) developed a basic comparison method for radiometrically calibrated data from the two aerial platforms based on regions of interests. We see this study as a starting point for a series of following studies to understand the environmental influence on UAS data and investigate the solutions to minimize such influence to ensure data quality.

Farmers can benefit from growing drought tolerant peanut (Arachis hypogaea L.) cultivars with improved yield when rainfall is sporadic. In the Virginia-Carolina (VC) region, drought is magnified by hot summers and usually occurs in July and Aug when pod and seed growth are intense. At these growth stages, weekly supply of 50 to 75 mm of water is needed to ensure profitability. Irrigation can supplement crop water needs, but only 10% of the peanut farms are irrigated.

In this frame, drought tolerant varieties can be profitable, but breeding for cultivars with improved drought tolerance requires fast yet accurate phenotyping. Our objective was to evaluate the potential of UAV remote sensing technologies for drought tolerance selection in peanut. In this study, we examined the effect of drought on leaf wilting, pod yield, grading characteristics, and crop value of 23 peanut cultivars (Virginia, Runner, and Valencia type). These varieties were arranged in a factorial design, with four replications drought stressed and two replications well-watered. Drought was imposed by covering the drought stressed plots with rainout shelters on July 19; they remained covered until August 29 and only received 38 mm irrigation in mid Aug. The well-watered plots continued to receive rain and supplemental irrigation as needed. During this time, Canopy Temperature Depression (CT) and Normalized Differential Vegetative Index (NDVI) were collected from the ground on all plots at weekly intervals. After the shelters were removed, these measurements were collected daily for approximately 2 weeks. At the same time, Red-Green-Blue (RGB), near-infrared (NIR), and infrared (IR) images taken from an UAV platform were also collected. Vegetation indices derived from the ground and aerial data were compared with leaf wilting, pod yield and crop value. Wilting, which is a common water stress symptom, was best estimated by NDVI and RGB, and least by CT; but CT was best in estimating yield, SMK and crop value in particular when taken on the ground at 15 days water stress imposition. Interestingly, CT predicted well plant wilting even before it occurred, i.e., correlation coefficients were negative and over 0.750 when CT was measured on July 19 and 20 even though wilting was visible only after two weeks. The data, yet preliminary, show promising potential for remote sensing technologies, at the ground and aerial, for peanut variety selection for improved drought tolerance.

Lodging has been recognized as one of the major destructive factors for crop quality and yield, particularly in corn. A variety of contributing causes, e.g. disease and/or pest, weather conditions, excessive nitrogen, and high plant density, may lead to lodging before harvesting season. Traditional lodging detection strategies mainly rely on ground data collection, which is insufficient in efficiency and accuracy. To address this problem, this research focuses on the use of unmanned aircraft systems (UAS) for automated detection of crop lodging. The study was conducted over an experimental corn field at the Texas A and M AgriLife Research and Extension Center at Corpus Christi, Texas, during the growing season of 2016. Nadir-view images of the corn field were taken by small UAS platforms equipped with consumer grade RGB and NIR cameras on a per week basis, enabling a timely observation of the plant growth. 3D structural information of the plants was reconstructed using structure-from-motion photogrammetry. The structural information was then applied to calculate crop height, and rates of growth. A lodging index for detecting corn lodging was proposed afterwards. Ground truth data of lodging was collected on a per row basis and used for fair assessment and tuning of the detection algorithm. Results show the UAS-measured height correlates well with the ground-measured height. More importantly, the lodging index can effectively reflect severity of corn lodging and yield after harvesting.

Grape vines are susceptible to many diseases. Routine scouting is critically important to keep vineyards in healthy condition. Currently, scouting relies on experienced farm workers to inspect acres of land while arduously filling out reports to document crop health conditions. This process is both labor and time consuming. Using drones to assist farm workers in scouting has great potential to improve the efficiency of vineyard management. Due to the complexity in grape farm disease detection, the drones are normally used to detect suspicious areas to help farm workers to prioritize scouting activities. Operations still rely heavily on humans for further inspection to be certain about the health conditions of the vines. This paper introduces an autonomous transition flight control method for a transformable drone, which is suitable for the future virtual presence of humans in further inspecting suspicious areas. The transformable drone adopts a tilt-rotor mechanism to automatically switch between hover and horizontal flight modes, following commands from virtual reality devices held in the ground control station. The conceptual design and transformation dynamics of the drone will be first discussed, followed by a model predictive control system developed to automatically control the transition flight. Simulation is also provided to show the effectiveness of the proposed control system.

Variety selection and seeding rate are two important choice that a peanut grower must make. High yielding varieties can increase profit with no additional input costs, while seeding rate often determines input cost a grower will incur from seed costs. The overall purpose of this study was to examine the effect that seeding rate has on different peanut varieties. With the advent of new UAV technology, we now have the possibility to use indices collected with the UAV to measure emergence, seeding rate, growth rate, and perhaps make yield predictions. This information could enable growers to make management decisions early in the season based on low plant populations due to poor emergence, and could be a useful tool for growers to use to estimate plant population and growth rate in order to help achieve desired crop stands. Red-Green-Blue (RGB) and near-infrared (NIR) images were collected from a UAV platform starting two weeks after planting and continued weekly for the next six weeks. Ground NDVI was also collected each time aerial images were collected. Vegetation indices were derived from both the RGB and NIR images. Greener area (GGA- the proportion of green pixels with a hue angle from 80° to 120°) and a* (the average red/green color of the image) were derived from the RGB images while Normalized Differential Vegetative Index (NDVI) was derived from NIR images. Aerial indices were successful in distinguishing seeding rates and determining emergence during the first few weeks after planting, but not later in the season. Meanwhile, these aerial indices are not an adequate predictor of yield in peanut at this point.

Millions of Americans come into contact with turfgrass on a daily basis. Often undervalued and seen as visual support stimulus for a larger entity, millions of acres of turfgrass can be found on residential lawns (which also provides an area for recreation), commercial landscape, parks, athletic fields, and golf courses. Besides these uses, turfgrass provides many functional benefits to the environment, such as reducing soil erosion, cooling its surrounding area, and soil carbon sequestration. However, rapidly expanding uses of turfgrass have also raised alarm for natural resources conservation and environmental quality, the largest impact being water consumption. This paper presents a machine learning approach that can assist growers and researchers in determining the overall quality and color rating of turfgrass, thereby assisting in turfgrass management including optimized irrigation water scheduling. Tools from Google and NVIDIA enable models to be trained using deep learning techniques on personal computers or on small form factor processors that can be used aboard small unmanned aerial vehicles (UAVs). The typical evaluation process is a long, laborious process, which is subjective by nature, and thus often exposed to criticism and concern. A computational approach to quality and color assessment will provide faster, accurate, and more consistent ratings, which in turn will help increase irrigation water use efficiency. The overall goal of the ongoing research is to use deep learning techniques and UAV imageries for the turfgrass quality and color assessment and help all the stakeholders to optimize water conservation.

Because unmanned air vehicles (UAVs) are emerging as an indispensable image acquisition platform in precision agriculture, it is vitally important that researchers understand how to optimize UAV camera payloads for analysis of surveyed areas. In this study, imagery captured by a Nikon RGB camera attached to a Precision Hawk Lancaster was used to survey an agricultural field from six different altitudes ranging from 45.72 m (150 ft.) to 121.92 m (400 ft.). After collecting imagery, two different software packages (MeshLab and AgiSoft) were used to measure predetermined reference objects within six three-dimensional (3-D) point clouds (one per altitude scenario). In-silico measurements were then compared to actual reference object measurements, as recorded with a tape measure. Deviations of in-silico measurements from actual measurements were recorded as Δx, Δy, and Δz. The average measurement deviation in each coordinate direction was then calculated for each of the six flight scenarios. Results from MeshLab vs. AgiSoft offered insight into the effectiveness of GPS-defined point cloud scaling in comparison to user-defined point cloud scaling. In three of the six flight scenarios flown, MeshLab's 3D imaging software (user-defined scale) was able to measure object dimensions from 50.8 to 76.2 cm (20-30 inches) with greater than 93% accuracy. The largest average deviation in any flight scenario from actual measurements was 14.77 cm (5.82 in.). Analysis of the point clouds in AgiSoft (GPS-defined scale) yielded even smaller Δx, Δy, and Δz than the MeshLab measurements in over 75% of the flight scenarios. The precisions of these results are satisfactory in a wide variety of precision agriculture applications focused on differentiating and identifying objects using remote imagery.

Texas A and M University has been operating a large-scale, UAV-based, agricultural remote-sensing research project since 2015. To use UAV-based images in agricultural production, many high-resolution images must be mosaicked together to create an image of an agricultural field. Two key difficulties to science-based utilization of such mosaics are geographic registration and radiometric calibration. In our current research project, image files are taken to the computer laboratory after the flight, and semi-manual pre-processing is implemented on the raw image data, including ortho-mosaicking and radiometric calibration. Ground control points (GCPs) are critical for high-quality geographic registration of images during mosaicking. Applications requiring accurate reflectance data also require radiometric-calibration references so that reflectance values of image objects can be calculated. We have developed a method for automated geographic registration and radiometric correction with targets that are installed semi-permanently at distributed locations around fields. The targets are a combination of black (≈5% reflectance), dark gray (≈20% reflectance), and light gray (≈40% reflectance) sections that provide for a transformation of pixel-value to reflectance in the dynamic range of crop fields. The exact spectral reflectance of each target is known, having been measured with a spectrophotometer. At the time of installation, each target is measured for position with a real-time kinematic GPS receiver to give its precise latitude and longitude. Automated location of the reference targets in the images is required for precise, automated, geographic registration; and automated calculation of the digital-number to reflectance transformation is required for automated radiometric calibration. To validate the system for radiometric calibration, a calibrated UAV-based image mosaic of a field was compared to a calibrated single image from a manned aircraft. Reflectance values in selected zones of each image were strongly linearly related, and the average error of UAV-mosaic reflectances was 3.4% in the red band, 1.9% in the green band, and 1.5% in the blue band. Based on these results, the proposed physical system and automated software for calibrating UAV mosaics show excellent promise.

Unmanned Aerial Systems (UASs) have been touted as being of great value for agriculture. However, they have not yet permeated the fabric of the agriculture research or agriculture production systems significantly. This paper seeks to explain why adoption is low relative to excitement and to provide some guidance on what is needed for UASs to be effective in agriculture research and agriculture production.

The application of unmanned autonomous vehicles (UAVs), or "drones", to generate data to support better decisions for agricultural management and farm operations is a relatively new technology that is now beginning to enter the market. This potentially disruptive technology is still in its infancy and must mature in ways that the current market cannot clearly foresee and probably does not fully understand. Major technical and regulatory hurdles must be overcome before the full potential of this remote sensing technology can be realized in agricultural applications. Further, and most importantly, buyers and sellers in today’s market must both gain a deeper understanding of the potential that this technology might achieve and the technical challenges that must be met before advances that will bring significant market value will be possible. A lack of understanding of some of the basic concepts of remote sensing can translate into poor decisions regarding the acquisition and deployment of UAVs in agriculture. This paper focuses on some of the details of remote sensing that few growers, and, indeed, few university researchers fully understand.

Methods of artificial intelligence are a good solution for weather phenomena forecasting. They allow to process a large amount of diverse data. Recirculation Neural Networks is implemented in the paper for the system of thunderstorm events prediction. Large amounts of experimental data from lightning sensors and electric field mills networks are received and analyzed. The average recognition accuracy of sensor signals is calculated. It is shown that Recirculation Neural Networks is a promising solution in the forecasting of thunderstorms and weather phenomena, characterized by the high efficiency of the recognition elements of the sensor signals, allows to compress images and highlight their characteristic features for subsequent recognition.