Source: RESONON, INC. submitted to
UAV HYPERSPECTRAL MAPPING SYSTEM FOR PRECISION AGRICULTURE
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
TERMINATED
Funding Source
Reporting Frequency
Annual
Accession No.
0206549
Grant No.
2006-33610-16811
Project No.
MONK-2006-00079
Proposal No.
2006-00079
Multistate No.
(N/A)
Program Code
8.13
Project Start Date
May 15, 2006
Project End Date
Jan 14, 2007
Grant Year
2006
Project Director
Moon, T. S.
Recipient Organization
RESONON, INC.
619 N. CHURCH #3
BOZEMAN,MT 59715
Performing Department
(N/A)
Non Technical Summary
The emergence of compact, high-performance imaging spectrometers and small, military-grade Uninhabited Aerial Vehicles (UAVs) has created an opportunity to develop an airborne system that provides affordable, timely, map-registered, hyperspectral data for precision agriculture. The purpose of the Phase I effort is to design an airborne hyperspectral imaging system for flight on small UAVs and customized for precision agriculture.
Animal Health Component
100%
Research Effort Categories
Basic
(N/A)
Applied
100%
Developmental
(N/A)
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
2055310202050%
2057210202050%
Goals / Objectives
During the Phase I effort we will develop a hyperspectral imaging system for small UAVs (payloads under 10 pounds) for precision agriculture. The sensor system will be designed to provide low-cost, geo-referenced, airborne hyperspectral imaging data that can be delivered with the timeliness and mapping accuracy required for directing the precise application of agro-chemicals to soils and plants. Hyperspectral imaging (HSI), also known as imaging spectroscopy, provides a detailed reflectance spectrum for every pixel in an image. This information can be used for determining plant health, locating weeds, and measuring the fuel load in a forest. The usefulness of this type of imaging data is greatly enhanced when it can be accurately geo-referenced in a timely fashion. In order to meet this goal, the following components will be integrated to form a complete UAV-HSI system: a UAV capable of being hand or rail launched, an Inertial Measurement Unit (IMU) and Global Positioning System (GPS) for geo-referencing data, a sensor payload consisting of HSI and single board computer, and a ground-based computer for data processing. After hardware integration, a prototype system will be used to collect geo-referenced hyperspectral data so that an evaluation of the system mapping accuracy can be made. Preparations for a flight demonstration on a military-grade UAV will also be made in collaboration with experts in precision agriculture. The HSI sensor design will need to strike the proper balance between spectral and spatial resolutions, aircraft speed and altitude, payload constraints, and the precision required for agricultural applications. Data from the UAVs IMU and GPS will provide the necessary geographic position and sensor orientation and will be collected synchronously with the image data. The onboard computer will control the sensor and store the data until it is transferred to a laptop computer after the UAV has landed. This data will be used to create geo-referenced hyperspectral images with information indicating plant and soil conditions. A discounted cash flow analysis in the context of precision agriculture suggests that the proposed system will be able to provide geo-referenced hyperspectral data for over 300 acres of land during one-hour of flight and provide an internal rate of return well in excess of a normal cost of capital. The system is expected to improve on currently available hyperspectral systems in agriculture, since it is faster than a vehicle mounted system and does not generate dust that obstructs the target. Compared to manned aircraft, the operational cost is far less and the logistics are simpler. While no single sensor system will fulfill all of the requirements of precision agriculture, a small UAV-borne HSI system can provide critical data with timeliness, low cost, and high resolution.
Project Methods
The Phase I effort is expected to show that it is feasible to obtain geo-rectified hyperspectral data from small UAV platforms. Several key components are required to form a complete UAV hyperspectral imaging system that can provide timely and cost-effective map referenced data. This includes a UAV capable of being hand or rail launched, an Inertial Measurement Unit (IMU), a Global Positioning System (GPS), an on-board computer, a power supply, a laptop for post-flight data downloading and processing and a Hyperspectral Imager (HSI) that meets UAV payload constraints and customized for agricultural applications. Tying the hardware components together will require software development for data acquisition, synchronization and post-flight geo-rectification. Spatial resolution of airborne hyperspectral data is strongly dependent on the UAV ground speed and internal camera frame rates. Due to bottlenecks associated with data transfer and storage limitations, the spatial resolution in the along-track direction of existing visible and near-infrared (VNIR) imaging spectrometer designs is very poor, and the spectral resolution far exceeds what is necessary for most agricultural applications. With appropriate optical redesigns and camera windowing strategies, overall spatial resolution can be increased by striking a better balance between along- and cross-track resolutions and decreasing spectral resolution. The optical system for a customized imaging spectrometer design will be modeled and optimized with Zemax ray-tracing software. A CMOS camera will be used to take advantage of its flexible data handling capabilities to reduce the amount of over sampling in the spectral regime and help boost camera frame rates. The IMU and GPS for small UAVs are already available but will require data synchronization with the HSI. The data will be stored on the onboard computer and transferred to a laptop after the UAV has landed. After submitting the original proposal Resonon has flown an existing HSI on a Rascal RC airplane and Manta UAV. These test flights have shown that the type of IMU and GPS used on the majority of small military-grade UAVs does not provide data with the necessary precision to meet the goal of 15 cm accuracy. One option is to develop an optical method to enhance the UAV autopilot and navigational system. Software for sensor control and data acquisition will be written for the onboard computer and for the ground station laptop to facilitate data archiving and geo-rectification. Geo-rectification software is presently being developed at Resonon and promises to streamline post-flight data processing over available commercial packages. This will make the system more cost effective and provide a timelier mapping product. Proof of concept experiments will be performed using an existing Resonon HSI and uncertainties in the geo-registration will be determined using laboratory mockups, computer simulations and possible flight tests. Designs for an optimized system will be completed and ready for fabrication and flight trials in Phase I. Plans for system improvement and optimization during the Phase II effort will be made.

Progress 05/15/06 to 01/14/07

Outputs
For the Phase I effort we designed and assessed the performance of a hyperspectral imaging system for small Unmanned Aerial Vehicles (UAV). The sensor design is optimized for crop condition analysis and is capable of providing economical map registered imagery in order to enable high value precision agriculture applications. Remotely sensed images from UAV's can provide a high spatial sampling because of their low flight altitudes. Multiple flights during a single growing season are also more practical thereby further improving crop monitoring and enabling rapid response to changing agricultural conditions. Although currently available sensors on UAVs can provide high spatial resolution imagery, they have very limited spectral information, typically only 3 or 4 broad spectral bands per pixel. Our proposed UAV hyperspectral imaging system can record 128 narrow spectral bands per pixel providing much more valuable information for crop condition assessment and still meet very tight payload limitations and price constraints. To our knowledge, this would be the first commercially available hyperspectral imaging system for small UAVs. Geo-referencing the image data to field coordinates is essential to enabling precision agriculture applications. To meet this objective, we integrated the necessary UAV hardware components with a hyperspectral sensor in order to geo-reference the imaging data. Flight tests were conducted to assess the potential mapping accuracy of this sensor package with promising results. Construction of the hyperspectral sensor and flight tests over an agricultural target of interest has been planned for Phase II. ACR, Inc., the manufacturer of a small, military-grade UAV, and agrarius LLC, a remote sensing company which presently uses UAVs for vineyard monitoring, have agreed to partner with Resonon during the Phase II effort. Resonon has also been contacted by other UAV manufacturers with interest in using hyperspectral imaging for precision agriculture and possibly forest fire monitoring. It is anticipated that interest from UAV manufacturers and remote sensing companies will increase as this commercial sector continues to grow and the performance characteristics of Resonon's UAV imaging systems become more widely known.

Impacts
Obtaining accurate, timely, and affordable information on plant and soil condition is essential for enabling high value precision agriculture methods. In response to this need, we designed a hyperspectral imaging system for small UAVs. The proposed system can scan large areas quickly and efficiently while providing high-resolution, multi-band information in a timely, cost-effective manner. Benefits will include improved yields because problems can be more efficiently addressed, and lower environmental impacts due to a decrease in agrochemicals by accurately distinguishing between areas of need and those not needing attention. Other benefits will include improved environmental monitoring, accurate mapping of noxious weeds, and pipeline monitoring.

Publications

  • No publications reported this period