It is an everyday miracle that plants, powered by sunlight, convert vast amounts of inorganic matter into living tissues. That miracle is the foundation of most life on earth. The success of the human species is owed in many ways to our ability to develop and cultivate edible plants, and as the human population steadily grows – soon to pass 8 billion – a vast army of agricultural producers, researchers, developers, and policy makers are on the front lines of securing and increasing the world’s food supply, which faces threats such as climate change, pests, emerging diseases, economic inequality, and political instability.
Dr. Daeun “Dana” Choi is part of this army, working at the Gulf Coast Research and Education Center (GCREC) in Wimauma, Florida, about 50 miles southeast of Tampa. She works there with other specialists in Precision Agriculture, an engineering specialty that integrates a variety of new technologies to improve the sustainability, productivity, and security of the food supply. Imagine a drone that flies over 100 acres of strawberries or tomatoes and collects image data that can be interpreted to determine nutrient deficiencies, plant stress, or specific diseases on a nearly plant-by-plant basis and then schedules and delivers the needed nutrients or pesticides using a variety of ground-based robotic technologies. Using this very precise approach, threats to a crop are detected early and treated with a minimum of chemical input, thus protecting the environment and helping to ensure profitability for producers.
Choi often works with combinations of sensors, robotics, and artificial intelligence. She has applied these technologies to a wide variety of crops, including tree crops such as oranges, apples, and peaches as well as ground crops like strawberries, tomatoes … and button mushrooms.
Mushrooms? Indeed. If you’ve never thought of mushrooms as an agricultural commodity, consider that Americans consume almost a billion pounds of mushrooms a year. Mushrooms are grown commercially in large trays kept in racks. In harvesting these small, delicate fungi, they are picked by hand, one by one, to ensure product quality, shelf-life, and appearance. This labor-intensive work is tedious, time-consuming, and expensive. Mechanized methods have tended toward devices that simply cut all the mushrooms on the tray, whether they are ready for harvest or not. The device is too rough to be used for the fresh market – grocery stores and restaurants – which is where the vast majority of mushrooms go and where they must be as clean and bruise-free as possible.
Choi brought her knowledge of machine vision, artificial intelligence, and robotics to the team that addressed this interesting problem. They developed a prototype device that uses machine vision to locate a mushroom, attach a gentle suction cup to it, and cut the stipe (the stem of a mushroom). Testing showed that imitating the manual technique by bending the mushroom to snap it off near the base was as effective as cutting, and it saved time and effort. Choi also worked on a team that developed a machine vision application to evaluate the maturity of individual mushrooms. This type of machine-vision recognition integrated with robotics has many useful applications, and it is a hallmark of many of Choi’s projects.
At the other end of the size scale, Choi has worked extensively with tree crops, including oranges and apples, exploring applications of machine vision, artificial intelligence, and robotics in many aspects of grove management and yield determination. In a significant project for citrus, Choi worked on a device that uses machine vision and artificial intelligence to count dropped fruit in orange groves. Premature fruit drop can indicate disease, especially a very serious disease called huanglongbing (HLB; formerly citrus greening), a bacterial disease of worldwide concern that has seriously damaged Florida’s citrus since it was first identified in the state in 2005. The bacteria reproduce inside the tree and block delivery of water and nutrients. The tree sickens, leaves become deformed, and fruit tends to drop before it is mature. Thus, the crop is reduced, and eventually, the tree dies. To find trees exhibiting HLB symptoms, growers make random surveys of hundreds of acres of trees and manually count and examine fruit on the ground. Choi worked with a truck-mounted device carried through a grove at slow speed that counts fallen oranges by analyzing images. Choi added to this the ability to analyze the images and discriminate between recently dropped and rotting fruit. This is a challenge for a machine because of factors such as variations in fruit color, fruit partially obscured by leaves or other fruit, and changing light conditions on the ground in a grove of trees.
Choi achieved this through sophisticated programming and a deep understanding of machine vision. The result was a device with 90% accuracy or greater and many times faster than a person when compared to manual counting. Using GPS, the device links images to their location in the grove, thus mapping fruit counts and providing the grower with precise locations of diseased or nutrient–deficient fruit, better yield predictions and timing for harvest, and scheduled treatments that are highly targeted and possibly automated. In the packing house, the device can examine fruit after harvest, discriminating between mechanical damage, HLB infection, or another important disease of citrus, citrus canker.
In Choi’s work with apples, she used an even wider array of technologies for an equally wide array of grove operations. In one project, Choi worked on a team that integrated vehicle-mounted lidar for precision spraying on apple trees with a GPS-based navigation system that corrected for uneven grove terrain. The combination can reduce chemical use by up to 15%, yielding savings in spraying costs and a reduction in environmental impact. Choi was also involved in work to develop robotic pruning of apple trees, one of the most labor-intensive aspects of grove management, which accounts for up to 20% of labor costs in a market with fewer farm workers and rising labor costs. However, robotic pruning presents many special challenges, such as utilizing machine vision to select pruning locations, handling and cutting branches, planning the pruner’s path among the trees, and avoiding obstacles. This requires integration of several sensing and robotic technologies, governed by complex software. Choi worked on several aspects of this, bringing it closer to a practical robotic pruner. Choi worked on another interesting application using drones and thermal imaging to pinpoint heating needs in an orchard, another grove operation that relies on laborious, time-consuming, and imprecise manual efforts.
Choi is now bringing her considerable technical skills and insight to the production of smaller fruits and vegetables. Choi’s current focus is tomatoes and strawberries. At GCREC, Choi is virtually surrounded by fields of these crops in one of Florida’s most intensely cultivated and productive agricultural areas. Florida is the top state for winter tomatoes and ranks second in production of strawberries. Together, these crops bring hundreds of millions of dollars to the state and its producers. Maintaining the productivity of these crops is vital to the livelihoods of thousands of Floridians who work directly in production and to the many workers and industries that use these products. Productivity is also key to keeping consumer prices low while maintaining profitability. It’s a delicate balance, and technology is key in maintaining it.
Many other specialty crops are grown in southwest Florida, and each one faces unique challenges. Choi works to give producers an advantage through the application of technology. Through her work, the latest developments in robotics, sensing technologies, and artificial intelligence are brought together to provide practical tools for producers.
Choi has said that one of the advantages of living in the area and working at GCREC is that she can get to know, and work directly with, producers. She can learn about and understand their concerns, which helps direct her research. At the same time, she can put a friendly face on new and possibly unfamiliar technologies, thus creating a more receptive audience for new methods.
How important is this work? Choi’s developments in precision agriculture have been praised for the use of “low-cost sensors,” which makes these methods more widely available and thus of benefit to some of world’s most important food crops. Choi’s extension of her work to small fruits and vegetables, which are critical in agricultural economies worldwide, extends the potential impact of this high tech – but accessible – technology for producers, both big and small.
That’s ABE: Big Questions, Global Reach!
This article was written by: Charles Brown