Helping Growers Care for the Apple Crop – They’re Working on an App for That

Image

AMHERST, Mass. – For apple growers in Massachusetts and across the country, thinning the crop each year is a labor-intensive practice that must be done to ensure that trees bear fruit each year, that there are not too few or too many fruits, and that mature apples are a good size for market, says Dan Cooley at the University of Massachusetts Amherst’s Stockbridge School of Agriculture and the School of Earth and Sustainability.

He adds, “It’s one of those jobs that is hard to get right. If you don’t thin fruit you’d get a lot of really small apples one year, and no fruit the next,” the professor of plant pathology explains. “It’s a balancing act that’s hard to get right. It depends on weather, the kind of trees you have, how healthy the trees are, and what you use to do the thinning. There’s a lot of uncertainty.”

“This is one of the steps in what is known in the industry as crop load management. It’s an ongoing lifetime-of-the-tree practice that must be done.” Some growers use chemicals to thin and some chemicals are worse for the environment than others, Cooley says, and using the right chemical at the right time is as much art as science, and is not clear-cut.

In a joint project with roboticists at Carnegie Mellon University in Pittsburgh, Cooley and extension fruit tree specialist Jon Clements at UMass Amherst have received about $430,000 of the two-year, $1.2 million grants from the USDA National Institute of Food and Agriculture and the National Science Foundation to develop a smartphone-camera-based system that will allow growers to quickly and accurately measure and thin fruit.

He and his team hope to develop a way that growers can use cell phone camera pictures to measure fruit quickly and easily. They want to create a system that reduces measuring time while at least maintaining – and potentially improving – accuracy, a method that is useful for a number of cultivars, cost-effective and easy for growers to use.

The research received support for preliminary studies from the Center for Agriculture, Food and the Environment and its Agricultural Experiment Station at Cold Spring Orchard Research and Education Center in Belchertown.

Cooley says this branch of study began years ago as Duane Greene, an apple thinning expert on the Stockbridge School faculty, developed a new way to make thinning more scientific and accurate. Greene’s fruitlet hand-counting and measuring system put these data through a computer-based model and it works well, but is tedious for growers.

“It’s a good prediction model, but to use it, growers have to mark about a hundred tiny fruitlets just after the apple petals fall, then measure them with a micrometer, then wait five or six days, and go back and measure each of the same fruitlets again to see whether they’ve grown and how much. It takes hours, and they tend not to bother,” Cooley says.

Cooley says robotics engineers led by Zack Rubenstein and George Kantor at Carnegie Mellon have had success in designing a robotic camera that is used in large vineyards for thinning now. But scaling it down to make a robotic “hand” that can measure fruitlets from the correct angles for use by small-scale growers will be a challenge.

Clements notes that it’s unclear whether automated apple growing will ever be possible, “but we’ll give this a try. Another thing we’re exploring is using precise weather data from several electronic weather stations in the orchard to give us a better picture of how sun, temperature, soil moisture and other factors are affecting tree growth and thinning.”

Cooley says, “I’m excited about it. The original idea was to keep it simple, put it on a grower’s cell phone, and make it easy for them to use the scientific thinning model. If cameras can recognize people in a crowd or guide cars, we hope we can get phone cameras to quickly identify and measure apples.”

Paul O’Connor, a doctoral student in the Stockbridge School, says, “There are a lot of questions to address, but we have a few years of data and I do think that machine learning may solve the problem in a way we can’t even imagine. We get excited around here about making it easier for farmers to grow apples.”