Four economists and an engineer walk into a research project, do a bunch of math and imagine a robot that costs $55,332.
That’s the highest price a grower could pay for an automated apple harvester and bring in the crop for the same cost as hiring people, according to a research team trying to put some economic context to the question growers are starting to ask themselves: How much should I pay for a robot?
Karina Gallardo, a Washington State University agricultural economist, presented the findings on potential robot prices and harvest profitability at the Washington State Tree Fruit Association Annual Meeting in December in Kennewick.
The work, based on a lot of assumptions at this point, does not attempt to tell growers how much a harvest robot is worth, said Jeff Luckstead, a fellow WSU agricultural economist who co-authored with Gallardo. For one thing, the price of the first commercialized automated harvester will likely be higher than $55,332.
“No company would sell a robot for that much,” Luckstead said.
The economists had no choice but to build their framework on assumptions because … well, no one sells a robotic harvester commercially yet. They’re all still in research and development.
The math only aims to tell growers what they could pay for a robot and be just as profitable as if hiring human pickers, as well as what factors have the most influence on the outcome of said equations. The fruit industry expects growers will need to know that someday soon, so five scientists from four universities did some high-level calculations to at least give the question some perspective.
Diane Charlton of Montana State University, Stephen Devadoss of Texas Tech University and Stavros Vougioukas, an ag engineer from the University of California, Davis, also co-authored.
They based their mathematics on a partially constructed multi-armed prototype picker in Vougioukas’ lab, which he is building as part of a National Robotics Initiative grant focused on coordinating the arms to maximize picking speed.
He did not build a full harvester, just enough to explore ways to balance the load of each arm so that the number of arms scales linearly; in other words, each additional arm offers an equal picking boost. If not, some arms would have to wait while the others pick. Platform-mounted human workers have the same problem, he said.
For the economic analysis, they assumed each arm picks an apple every 3 seconds and the arms are perfectly synchronized to avoid delays. They also assumed the robot damaged 5 percent of the fruit, left 10 percent of the target fruit on the trees and would last 10 years. For the orchard, they assumed 90 acres of Galas where the base wage for humans is $18.79 per hour. They also assumed human pickers would take 120 worker-hours to harvest 1 acre while the robot would take about 14 hours.
In the end, the researchers determined the imaginary grower could pay $461.10 per month over 10 years — $55,332 in total — and profit the same as if they hired manual labor to pick those 90 acres of Galas.
Knowing things change all the time in farming, the researchers threw in some variables, including picking speed, wages and fruit damage. For example, they crunched the numbers with a range of single-arm picking speeds of 1 second to 4 seconds per fruit.
“Four seconds is an optimistic but not unrealistic number, and such top (not average) speeds have been reported in literature and by companies,” Vougioukas said. “One second is very optimistic, but we wanted to explore it.”
Among their findings, they determined that for every 1 percent increase in wages (say from $18.79 to $18.98 per hour) the robot would still break even if it:
—Picked 6.69 percent slower,
—Damaged 4.37 percent more fruit, or
—Left 0.22 percent more fruit on the trees.
Those are pretty big swings, Gallardo said.
“Results are very sensitive to small changes in the robot parameters, therefore the wide range of variation,” she said.
—by Ross Courtney