By Jamie Martin
Researchers at the University of Florida have created a digital twin of a strawberry field to support year-round agricultural innovation.
This virtual replica simulates every leaf, row, and berry, allowing scientists to test robotics and artificial intelligence tools, even outside the typical November to April harvest season.
A digital twin is a computer-simulated environment that mimics real-life systems. In this case, it has helped researchers develop a robotic strawberry-harvesting system without the need for real field data.
The team, led by assistant professor Dana Choi, used the model to train AI tools to detect and measure strawberries with 92% accuracy using only synthetic data.
“Because the computer-simulated field never goes out of season, new berry-spotting tools can be prototyped even in the summer – speeding innovation,” said Choi.
The research found that robots trained with this method estimated fruit diameter with an error of just 1.2 millimetres—accurate enough for commercial grading.
This eliminates the need to capture thousands of real photos and label them manually, saving time and reducing development costs.
The digital twin can also help with yield prediction by estimating fruit size and volume, enabling growers to plan better harvests. It supports operator training and rapid prototyping of autonomous machines, speeding up the journey from concept to field use.
For Florida’s $500 million strawberry industry—and the larger $2 billion U.S. market—these advancements could reduce reliance on manual labour and costly field trials.
With this realistic virtual platform, researchers can develop smarter farming tools quicker and more efficiently, bringing innovation to farms faster and more affordably.
Categories: National