How AI is Improving Food Security in a Warming, Growing World

Climate change, a growing population, and shifts in the labor market all jeopardize our ability to feed ourselves in the coming decades. Currently the gap between the expected gains in agricultural productivity and the necessary gains to sustain humanity is projected to widen considerably.

Artificial intelligence and robotics offer some tools to address this shortfall. In this discussion, two professors of the University of Florida--Carlos "Charlie" Messina and Changying “Charlie” Li describe their work on berries, corn, and other crops.

Sensing and Analyzing

Although they don't currently work together, the types of work the two Charlies do is complimentary. Messina uses AI to map key aspects of plants (phenotype), such as their productivity, to different genetic variations (genotype) that correspond to those attributes. With this data, and powerful algorithms, he is able to virtually breed scores of different plant varieties in an effort to identify with the most productive and resilient strains. Messina used these methods to improve the drought tolerance of corn (maize) while working for DuPont. He's now applying the technology to breed more productive strawberries, among other projects.

Charlie Li collects the kinds of phenotype data that can inform Messina's work. He's developed a modular, data gathering robot, called MARS, that can move through fields to collect visual information about the health and productivity of plants, as well as use depth sensors to understand their physical attributes. Li combines this information with other sensing technologies, such as LIDAR and hyperspectral sensors (working in near-infrared) to study attributes of plants and individual fruits.

Breeding Better Blueberries

Li's research has primarily been on blueberries, but it is transferable to other crop types. In addition to helping improve yields, Li is also studying ways to improve robotic harvesting, which he sees as critical as fewer workers show interest in the low wages, hard labor, and erratic schedules of harvesting jobs. Li is studying the best shape of plants to facilitate mechanical harvesting, as well as ways to minimize bruising of fruit that can spoil parts of the harvest. For the latter, he has developed a blueberry-shaped sensor to measure pressure and stress of harvesting machines. In other words, a robotic blueberry is assessing the work of a robotic farm hand. You can read more about Charlie Li's research in the Fall issue of Worth magazine.