Israeli agricultural robot aims to help farmers assess fruit yields

"Agriculture is the sole industry in which you have a production line that you don't know what the output will be at the end of it.”

DECEMBER 4, 2016 16:33

Updated: DECEMBER 4, 2016 17:46

Robotic sonar (Video: Avital Bechar)

Farmers of the not-so-distant future may be able to accurately project their fruit yields with the help of an automated “AGRYbot” currently taking shape in central Israel.

Known more formally as the “Robotic Sonar for Yield Assessment and Plant Status Evaluation,” the AGRYbot is a sonar system mounted on the end of a robotic manipulator that is capable of identifying the acoustic signature of different entities in the agricultural plot.

Employing sophisticated algorithms and sending out sound waves reflected in the environment, the AGRYbot will provide farmers with the number of leaves on plant branches, an indicator of potential fruit yield.

“We can distinguish between boxes and the infrastructure of the environment, the walls of the greenhouse, crop rows, etc.,” its chief inventor, Dr. Avital Bechar, a senior scientist at the Agricultural Research Organization in the Agriculture Ministry’s Beit Dagan Volcani Center, told The Jerusalem Post on Sunday.

“If you look at the plant with the sonar, we can distinguish between the plant, the leaves and the fruit themselves,” he continued.

The AGRYbot, Bechar explained, can calculate the weight of fruits on the plants it is scanning with an accuracy of about 100 grams, and determine the number of leaves present with an accuracy of about 30 leaves.

At the Agricultural Research Organization, Bechar serves as head of the Production, Growing and Environmental Engineering Department within the Institute of Agricultural Engineering. The AGRYbot is a joint project between his team and researchers at Tel Aviv University.

One big advantage of employing sonar to determine the number of leaves on a plant, he said, is that the technology is able to penetrate the foliage.

“Nowadays, farmers evaluate the yield by the naked eye with a small sample,” he said, noting that they typically analyze a few trees and then extrapolate results for their entire plot.

“This is not very accurate,” Bechar continued. “You can get errors of up to 50%. You don’t see the hidden fruit, and not all the trees and plants are similar. If you make an assumption on a very small sample that doesn’t represent the plot, you can get large errors.”

Having accurate yield assessments is a critical farming tool for a variety of purposes. For example, Bechar explained, one farmer might want to remove some of his or her fruit to ensure that the plants have fewer, but larger produce.

“Sometimes you remove fruit when you don’t need to, or you remove too few,” he said.

Yield assessments also enable farmers to create work programs and understand how many people need to be hired to harvest their fields, Bechar added.

“Agriculture is the sole industry in which you have a production line that you don’t know what the output will be at the end of it,” he said.

Bechar hopes the technology he and his team are developing will change the face of farming in the future. At the moment, however, they have a working prototype that operates both indoors and outdoors, conducting experiments.

The researchers presented the AGRYbot, which is expected to reach the market in three to five years, on Sunday to a Chinese business delegation led by Ronnie Chan, chairman of Hang Lung Properties Ltd. Members of the group viewed a number of agricultural innovation projects under development at the Volcani Center as part of their visit.

“The next stage will be to continue developing the sonar system and finding the party that will take this technology we developed and turn it into a product,” Bechar said. “I think the market is eager for such a solution for the yield-evaluation problem.