Brave new world of drones, ag data
FARGO, N.D. -- If machines are going to go out and identify weeds and then send out drones or sprayers to control them, the cameras researchers first must "teach" the machines how to tell the weeds from crops and weeds from weeds.
FARGO, N.D. - If machines are going to go out and identify weeds and then send out drones or sprayers to control them, the cameras researchers first must "teach" the machines how to tell the weeds from crops and weeds from weeds.
And if cattle farmers and ranchers are going to use drones to count cattle and check them for diseases, they'll need some "machine learning" to sort out what behavior is meaningless and what might indicate disease stress.
These are among the goals of new precision agriculture studies underway at North Dakota State University, which has just launched a new precision agriculture degree program with major and minor degrees precision agriculture.
Greg Lardy, NDSU associate vice president for agricultural affairs and interim director of Extension, is the principal investigator for a five-year project that runs through Aug. 31, 2023. The project was initiated by Dr. Sreekala Bajwa, who was the former chairman of NDSU Agricultural and Biosystems Engineering, but left to be vice president of agriculture at Montana State University. It is funded by a $4.3 million from the U.S. Department of Agriculture. NDSU's role is to conduct precision agriculture research, including UAV research and crop management technology. One is on precision weed management and another is on genetic environment and management.
NDSU's first-year funding is $870,000. Subsequent years are controlled by annual funding within USDA. The project is part of a larger project with USDA scientists and with Mississippi State University.
One of the co-investigators is John Nowatzki, an NDSU Extension machine systems specialist. Another is Xin "Rex" Sun, an agricultural engineer and assistant professor in NDSU's Department Agriculture and Biosystems Engineering.
Sun, a doctor of agricultural engineering, is one of the professors who in the 2019 spring semester started for the department's Precision Agriculture major and minor degrees. In addition to crops, he also works in livestock precision agriculture.
One of Sun's main crop research projects involves using ground-sensing optical sensors to develop an algorithm to identify different weed species. The project starts in the greenhouse, and acquiring detailed data on growth, shape and other characteristics of weeds.
The first step in the project is to distinguish weed species among each other and from soybeans. Later, the weeds will be distinguished from corn and wheat crops.
The group has built a ground sensor physical platform in the greenhouse. The equipment uses "hyperspectral" imaging and three-dimensional imaging to focus on key weed species.
The machines began taking pictures in April. In June, some of the research will be taken to the field, using a ground sensing information from a drone.
One goal is to take the information to the field and take an autonomous vehicle to the field-a "smart sprayer," to target-spray weed-killing herbicides. The trick is to eliminate the weeds under 5 inches tall.
The researchers use RGB (the humanly visible, red-green-blue wavelengths) as well hyper-spectral imaging (including infrared) technology from a drone that can be used with wide-range sprayers, turning nozzles on and off based on what the drone sees.
Some plants may be distinguished with these technologies even if the differences aren't obvious to the naked eye.
In addition to precision crop production, Sun also brings his precision ag expertise to livestock studies, including using:
• Hand-held thermal imaging device to detect early-stage cattle respiratory diseases. The thermal imaging map will be correlated with rectal temperatures, which are the typical indicator of respiratory problems.
• Video and thermal imaging processing technology to evaluate the relationship between beef cattle temperament and beef meat flavor and quality. Theoretically, calmer cattle produce better beef quality. Historically, temperament is measured using the human eye. "That's highly dependent on the human experience," Sun says.
• Thermal imaging, near-infrared to determine cattle temperament data as an indicator of respiratory disease. Essentially, an animal with respiratory discomfort moves differently than cattle that are healthy.
• Drones to count beef cattle on a pasture. In the future, the project may be able to help detect behavior related to cattle respiratory diseases, or even identify weeds in a pasture. This is a pilot study priority for the State Board of Animal Research and Education.