Imagine a sea of wild berry bushes shimmering in the sun, the day’s harvest nearly done. Farm workers pack up after long hours in the summer heat, leaving behind berries that have ripened too late to pick, but as the field quiets, something shuffles busily between the rows — a small robot designed to keep working through day and night, gently plucking the fruit human hands did not reach.
The economics of berries in the U.S. are significant: U.S. fruit, tree nut and berry farms reported $34.2 billion in sales in 2022, and hired-labor expenses on berry/fruit specialization farms were $7.6 billion, 26 percent of production costs. Meanwhile, the labor supply that underpins hand-harvested crops is shrinking.
New-entrant crop-farm workers now represent only about 3.6 percent of the hired crop labor force, down from around 22 percent during the late 1990s. For labor-intensive crops, reductions in labor supply translate into real losses: One study found a 10 percent decline in labor supply could reduce production by roughly 4.2 percent in key U.S. hand-harvested regions. For farms where fruit must be picked by hand to avoid bruising and value loss, a dwindling workforce is not just a cost issue; it is a threat to harvest volume, grade and, ultimately, price.
Anthony Gunderman, an assistant professor of mechanical engineering at the University of Arkansas in Fayetteville, is an Arkansas native. He grew up in Batesville, where he was inspired to pursue engineering at UA by his parents, both of whom are electrical engineers. Gunderman eventually arrived at Georgia Tech, where he nurtured his curiosity in robotics. When he accepted a position at the UA, he reconnected with a pair of colleagues, Renee Threlfall and Amanda McWhirt, who were destined to reintroduce him to the Natural State’s multimillion-dollar berry industry.
However, Gunderman’s agricultural exploration started during his graduate degree under the supervision of his then-advisor, Yue Chen.
“Dr. Chen had wanted to use soft robotic grippers in agriculture, and he was looking for a practical use,” Gunderman said. “Dr. Renee Threlfall, who’s a food scientist at UA, realized that this type of device could solve a big problem with harvesting fresh-market blackberries.”
Blackberries have a 30-day harvesting window, and they ripen in a staggered fashion. One variety might ripen early, one variety might ripen later, so growers and laborers frequently revisit plants to determine the time of harvest. It is a costly allocation of labor, but what if farmers could automate the process without damaging the fruit?
“That’s where the soft-touch robotic element comes into play,” Gunderman said, holding a three-pronged soft robotic gripper made of silicon and smaller than a human hand. The device was created by Gunderman and Chen while Gunderman was a Ph.D. candidate in Chen’s lab.
The challenge extends beyond creating what is arguably the best gripper for picking fruit, however. The robot must also distinguish between a ripened and nonripened berry, traditionally an exclusively human skill. Gunderman and his team are exploring a myriad of robotically measurable parameters to determine berry ripeness, including berry detachment force, infrared imaging and camera images. That exploration is aided by an Arkansas Research Alliance Impact Grant.
“To determine what makes a berry ripe, we picked and measured nearly 2,000 berries,” Gunderman said. “Each berry was first imaged, provided a unique serial number and harvested using a custom-designed detachment-force-measuring gun. Each berry was then brought back to the lab for soluble solids and pH analysis.
“Our colleague, Dr. Dongyi Wang, performed near-infrared reflectance analysis so that we could identify which wavelength a ripe blackberry reflects most strongly. Dr. Ngan Le and Dr. Chase Rainwater are analyzing camera images of berries to see if images provide enough information to detect the nuance of berry ripeness.”
Artificial intelligence also plays a crucial role in the robot’s development, serving as a necessary tool to analyze the camera and near-infrared images and answer the question, “Is this berry in this image ripe, or is it not?”
The process of developing a berry-picking robot has given Gunderman even greater respect for growers and laborers.
“Humans are incredibly fast at picking berries,” he said. “On our first harvesting day, we only harvested 60 berries over eight hours. However, that helped us better understand the hurdles we had to overcome. How do you image the berry first? What is your protocol for that? How do we harvest the berry in an efficient manner? How do we store them? How do we get them back to the lab? It was interesting to me to see how many hurdles we had to overcome after that first day, but even as we optimized our workflow, we were only able to harvest about 700 berries over the course of five to six hours.”
No matter how smart the technology becomes, Gunderman said he does not see robotics replacing laborers.
“Ideally, I see robots as a tool that works alongside pickers. The main objective is just to augment the harvesting capacity so that you maximize yield because fruit that gets knocked off or picked by birds reduces profit. After seeing how fast humans can harvest, I’m not sure we’ll ever be able to match that efficiency, but we can develop multiple robotic systems that can harvest all throughout the night and day, providing consistent harvesting capability.”
Should Gunderman’s project reach its fullest potential by becoming a fully automated and effective agricultural tool, what ultimate benefits would his research achieve?
“Our hope is that over the next five years, we walk away with a robotic system that can either harvest autonomously or can come just short of that,” he said. “The cool thing about this is that if it works for blackberries, there’s no reason we can’t extend that to raspberries and other types of fruit. Long term, I’d like to see some sort of franchise-based system or a company that leases the use of that robot to the growers to augment their labor force in a reliable way that maximizes yield.”
A franchise like that born from Arkansas research could create new high-tech jobs, attract investment and open new opportunities for value-added innovation across the state’s agricultural economy. It is a reminder that innovation does not just happen in labs; it starts in the field, where problems meet curiosity. The ground floor for building Arkansas’ next generation of high-tech companies begins with research like that led by Gunderman and supported by the Arkansas Research Alliance.
Photo provided by the University of Arkansas
READ ALSO: Kyron Jones Named CEO of Academics Plus Charter School
