Picking lettuce by hand can be a labor-intensive and uncomfortable job for farmhands, as the crop is low to the ground and easily damaged.
And laborers to perform that job are getting harder and more expensive to find.
Enter the “Vegebot,” a robot trained to pick iceberg lettuce, a crop that so far has been difficult to harvest using automation.
The Vegebot features a computer vision system and a cutting system. An overhead camera takes an image of the lettuce field and, using artificial intelligence techniques, identifies which heads are ready for harvesting and which are unripe, damaged, or diseased. A second camera is located near the blade to help get the most precise cut for the ripe lettuce.
“The biggest challenges in the development of the Vegebot were developing a computer vision system to locate the lettuce, as the fields look like a sea of green leaves to the untrained eye, and picking the lettuce reliably without damaging it,” says Simon Birrell, a research student focused on machine intelligence at the University of Cambridge’s engineering department.
Birrell and his colleagues at the university co-authored a paper about the Vegebot published in the July 7 issue of The Journal of Field Robotics.
Agri-tech, or automation devices used for agriculture, already are in wide use in food packing facilities, but they are becoming ever more commonplace in the field with spraying, planting, and now, picking machines, says Ryan Jacobsen, CEO of the Fresno County Farm Bureau, Calif., a nonprofit advocate for farmers representing 3,000 members. The bureau is part of the nationwide group of farm bureaus in every state with a total of five million members.
“We are at an incredible time with automation and technology being deployed rapidly on farms in California. I expect that to increase in the next five years,” he says. “The plusses of automation are for labor efficiency and for the precision of the equipment. The accuracy is pretty incredible.”
Field workers are getting more difficult to find, he says, and they are expensive. And while robots aren’t ready to replace humans, they could work side-by-side and help improve picking efficiency in the field, he says.
In choosing lettuce for their robot to pick, the University of Cambridge researchers hit on one of the challenging points in agri-tech.
“Lettuce is extraordinarily labor intensive to pick,” Jacobsen says. It is low to the ground and delicate.
Adds Jacobsen, “And agriculture is one of the more dangerous professions in the country. So any time we can find ways to improve the safety and comfort for people it’s good.”
The Vegebot initially was trained to recognize and harvest iceberg lettuce in a lab setting. But the scientists have successfully tested it in a variety of field conditions in cooperation with G’s Growers, a U.K. fruit and vegetable co-operative.
Their prototype still is much slower and less efficient than a human worker, but it is a step forward in automating iceberg lettuce harvesting. The robot also may help reduce food waste by allowing more times to pick lettuce.
Potatoes and wheat are among the crops that have been harvested mechanically at scale for decades, but iceberg lettuce, which is the most common lettuce grown in the U.K., is easily damaged and grows relatively flat to the ground, presenting challenges for robotic harvesters.
“Every field is different, every lettuce is different,” Birrell says. “But if we can make a robotic harvester work with iceberg lettuce, we could also make it work with many other crops.”
Harvesting is the only part of the lettuce life cycle that is done manually, says co-author Julia Cai, who worked on the computer vision components of the Vegebot.
In addition to identifying which lettuce to pick, the researchers were also able to adjust the pressure in the robot’s gripping arm so that it held the lettuce firmly enough not to drop it, but not crush it. The force of the grip can be adjusted for other crops.
“We wanted to develop approaches that weren’t necessarily specific to iceberg lettuce, so that they can be used for other types of above-ground crops,” says Fumiya Iida, PhD, who leads the team behind the research.
Robotic harvesters also could be trained to harvest around the clock. Currently, each field is harvested once and unripe food or vegetables are discarded. A robot could be trained to pick only ripe vegetables and perform multiple passes on the same field on different days to harvest produce that ripens later.
Birrell says the researchers are working on a project to have the Vegebot learn so it can improve its performance as it operates.
“Beyond that, we’re interested in working with commercial companies who would like to turn it into a product for the market,” he says.