Source article published on November 06, 2023
Harvesting the Future: AI-Driven Robots Transforming Specialty Crop Farming
Specialty crop growers face considerable labor expenses in agriculture, yet the delicacy of the produce makes automation challenging.
Harvesting blackberries presents a complex task, whether humans are part of the process or not.
The fruits need to be plucked right away to prevent deteriorating and turning into a tasty treat for hungry animals. In contrast to other crops, growers must handpick the berries since heavy gear cannot handle the fruit’s fragile structure.
Dr. Renee Threlfall, a food scientist at the University of Arkansas, stated that blackberries are a delicate and tender fruit that must be hand-picked when they are shiny and black in color.
Growing fresh fruit is getting harder and more expensive for growers of specialist crops like blackberries, as labor prices rise quickly and the workforce shrinks. An autonomous, robotic hand is what researchers at Georgia Tech and the University of Arkansas are hoping to offer the solution.
Empowering fruit farming with AI-driven robots
A completely automated harvesting system that is sensitive enough to select fruits is being developed by Threlfall and others. Once developed, this technology could aid farmers in the specialty crop industry. Such farmers, often overlooked, stand to benefit by boosting their yields and thus reducing operational costs.
According to Threlfall, the scarcity of manpower to gather fresh-market blackberries is a significant obstacle. Therefore, addressing labor challenges has the potential to boost U.S. blackberry production.
Threlfall collaborated with researchers in Arkansas and Georgia in 2020. This partnership included Professor Yue Chen, an assistant professor in the Department of Biomedical Engineering at Georgia Tech and Emory University.
Their objective was to develop an autonomous robot capable of picking berries with same gentleness and precision as human. The team has recently obtained grant funding from the National Institute of Food and Agriculture’s Cyber-Physical Systems program. This collaboration is in conjunction with the National Science Foundation.
From Surgery to Harvesting: Innovations in Robotic Fruit Picking
Chen is a biomedical engineering specialist that focuses on robotic surgery that is less invasive. Therefore, he discovered that plucking fragile fruit is comparable to doing delicate surgery.
According to him, “both depend on precise motion planning, robot control, navigation, and end effector manipulation. So, I applied the concepts of medical robotics to agriculture.”
The research process commenced with a few fundamental inquiries: when harvesting blackberries, how many fingers does a person need? How much pressure is necessary, and at what point will the fruit be harmed?
The researchers utilized silicone finger wraps equipped with biometric sensors to monitor force and movement. Graduate students harvested hundreds of blackberries to collect valuable data. The information showed that they use only three fingers and exert about 0.5 newtons of force while picking.
Based on these discoveries, Chen and his doctoral student Anthony Gunderman created a robotic system that picks berries 95% of the time with enough gentleness to avoid damage. This success rate surpasses that achieved by the majority of human pickers.
Equipping the hand with a mind
Teaching the robot which berries were ready to be harvested was the next challenge. In essence, they had to load it with artificial intelligence. Specialty crop growers are just now beginning to deal with it. However, there is no straightforward solution in sight.
Computer vision, a type of AI, can examine the branch structures to decide which fruits to pick. Yet, environmental factors and the expertise demanded for robot operation remain significant challenges. These factors collectively hinder the wider implementation of automation within the specialty crop industry.
“Using cameras to perceive fruit and branches becomes more difficult in open-ended environments like berry bushes,” stated Alan Fern, an Oregon State University computer science professor and co-principal investigator at the AgAID Institute, a joint initiative to use artificial intelligence to solve challenging agricultural problems. Furthermore, operating robot manipulators calls for extreme dexterity and little margin for error.
For Chen, the key hurdle was acquiring a sufficiently extensive database to educate the robot on identifying ripe berries and their locations.
