Precision farming techniques for citrus production

Precision farming techniques for citrus production

Precision farming has been a big topic in agriculture as it maximizes yields and reduces input costs for growers. A UF/IFAS research group is currently conducting a variety of different research projects on applying precision farming techniques to a multitude of different areas in citrus. Dr. Yiannis Ampatzidis, an assistant professor who leads a research and extension precision agriculture engineering program in the Southwest Florida Research and Education Center (SWFREC) in Immokalee, presented on the research thus far earlier this year. 

Using Precision Farming Techniques

Ampatzidis shared how the different research projects are utilizing the techniques of precision farming—using Unmanned Aerial Vehicles, or UAVs, also referred to as drones—to collect data and then using cloud-based software to make the data useful by processing that data, analyzing it, and creating a way to visualize it. 

The team is utilizing technology such as thermal cameras, Multi-Spectral imaging, Visual – RGB, and LiDAR (3-D laser scanning) coupled with drones and Artificial Intelligence (AI) to conduct their research. According to Ampatzidis’ presentation, the research is taking advantage of AI and a deep learning algorithm, and then using existing neural networks such as Alexnet and Googlenet to “train them to identify and detect objects according to our requirements.”

For instance, one neural network is being trained to detect and identify citrus flowers, leaves, and fruit, and then categorizing those things as either “healthy” or “unhealthy.” Once identifications and categorizations are made, a map can be created that represents the health of the entire grove down to each individual tree. Ampatzidis says they are planning on utilizing similar technology on equipment like harvesters and weed blasters to improve fruit harvesting and weed elimination.

The data collected can be sorted into visual representations, from maps to graphs, so citrus growers can identify healthy or unhealthy trees, areas of the grove that are underproducing, the presence of pests, and more. One project is focused on gaining data on the tree heights, tree canopy area, and tree health of a grove section. Another project is analyzing rootstocks. A third counts fruit on each tree in the grove. There’s even a project that utilizes a robotic arm on a utility vehicle that will conduct tap samples on citrus trees for Asian citrus psyllids, count the psyllids, and then create a visual map identifying the levels of infestation in all areas of the grove.

All of these projects are aimed at achieving the same end result: taking money- and time-consuming tasks that citrus growers have heretofore had to complete by hand, and letting UAVs, AI, and software do the hard and tedious work instead.