How AI and Unmanned Aerial Systems Could Change the Future of Crop Scouting
Crop scouting may transition from a boots-on-the-ground job to an artificial intelligence endeavor in the sky thanks to research from The Ohio State University (OSU) and investments made by the Ohio Soybean Council (OSC) and soybean checkoff. Dr. Scott Shearer, professor and chair of OSU’s Department of Food, Agricultural and Biological Engineering, and his team are testing the use of small Unmanned Aerial Systems (sUAS) in Ohio fields to automate the scouting process with data collected directly from the crop canopy.
To dig deeper, OSC talked with Dr. Shearer about the project and the impact it could have on Ohio agriculture.
Q: Tell us about your current work with AI and sUAS.
A: We have developed a stinger platform suspended beneath a multi-rotor drone, or sUAS, to insert sensors into the crop canopy. These sensors capture high-resolution imagery from within the plant canopy, which can be used for real-time plant stress classification.
Over the past two growing seasons, we have been scouting soybean fields and building an extensive image library of soybean crop stress imagery. Convolutional Neural Networks (CNNs), AI algorithms used for image recognition, have been trained using the image library to support real-time classification of crop stress. The resulting CNN classifiers are being field tested for accuracy.
Currently, the predominant sensing technique uses low-cost RGB cameras. However, additional work has been conducted this growing season to include a near-infrared spectroscopic sensor as well as tissue sampler, both suspended on the stinger beneath the drone.
Q: How does this technology benefit Ohio soybean farmers?
A: The direct benefit to Ohio soybean farmers is a more efficient and accurate scouting approach for improved crop health monitoring. Current scouting practices require the farmer to scout three or more locations within a field. However, the sUAS approach significantly expands the scout’s ability to monitor many more sites within a field and to automate the stress detection and specification process using AI.
Ideally, farmers using this method will be alerted of stressors affecting their soybean crop sooner so they can implement corrective measures more quickly and preserve yield potential for improved profitability. This rapid assessment approach will move the industry toward a more prescriptive approach to crop stress management where economic thresholds are addressed on a refined spatial basis.
Q: When do you anticipate this technology could be commercialized?
A: Researchers and technology commercialization managers have been in contact with several venture capitalists and ag tech providers to explore commercialization options. The approach is somewhat constrained by the ability to develop regional and crop-specific reference libraries to train the CNN classifiers, so the value of this approach depends on who can develop those libraries. It’s likely the first commercial deployment of this system will occur within three to five years.
This technology may be a few years from your farm, but there are many other innovations coming available every day to help improve the efficiency of your operation. Find out which technologies are currently working for other Ohio soybean farmers here.