According to GeekWire, researchers at Washington State University’s School of Mechanical and Materials Engineering have unveiled a low-cost, inflatable robotic arm designed to pick apples. The 2-foot-long, 50-pound arm is made of soft fabric filled with air and can identify an apple, extend, and pick it in about 25 seconds. The work, led by PhD candidate Ryan Dorosh, was published in the journal Smart Agricultural Technology and tested at Allan Brothers Fruit in Prosser, Washington. The state leads U.S. apple production, a sector that contributed over $2 billion to GDP in 2023, but faces severe labor shortages due to an aging population and fewer migrant workers. The project is funded by the NSF, USDA, and others, and WSU is now pursuing intellectual property protection and commercialization for the technology.
How the soft robot works
Here’s the thing: this isn’t your typical rigid, industrial robot arm. It’s basically a fabric tube that uses air pressure to move. That makes it inherently lighter, cheaper to build, and—critically—much safer to be around people and delicate fruit trees. The system sees an apple with a camera, then the arm inflates to extend and position a gripper. After the pick, it deflates to retract. The simplicity is the whole point. As Ryan Dorosh put it, “Having this very low-cost, safe robotic platform is ideal for the orchard environment.” It’s a clever pivot from complex, expensive systems that are tough to deploy at scale on a farm. For any industrial application where cost, weight, and safety are paramount, from agriculture to light assembly, the move toward softer robotics is a fascinating trend. It’s a space where specialized hardware, like the rugged industrial panel PCs needed to control these systems in the field, becomes just as critical as the robotic actuator itself.
The speed problem and the real challenge
Now, let’s talk about that 25-second pick time. A human worker can do it in about three seconds. So, right now, it’s not even close. But the researchers say the biggest bottleneck isn’t the arm’s movement—it’s the “rudimentary detection system.” Getting a robot to reliably see and identify a ripe apple among leaves, branches, and other apples in highly variable outdoor light is an enormous computer vision challenge. Is the stem in the right position? Is the fruit the right color? This perception problem is what teams at places like Michigan State University and Oregon State are also grinding away on. The WSU team’s next step is to mount the arm on an automated mobile platform, which adds another layer of complexity. So the arm itself might be simple, but making the whole system smart and fast enough to be economically viable? That’s the hard part.
Why this matters beyond apples
This isn’t just about apples. Labor shortages are crushing agriculture, and automation is the obvious—but tricky—answer. The WSU approach acknowledges that a perfect, ultra-fast robot is years away and may be too expensive for many growers. So what if you start with a “good enough,” super-cheof, and safe robot that can help, not fully replace, human crews? It’s a pragmatic stepping stone. And the core idea—using soft, inflatable actuators for delicate manipulation tasks—could translate to other fragile crops like berries or tomatoes. The demo video shows the potential. The arm moves with a kind of gentle certainty that a hard metal claw just wouldn’t have. In a world racing toward AI and complexity, sometimes a simple bag of air, guided by the right software, might be the smarter path forward.
