Researchers at Chinese universities have developed BioflexBot, a robot gripper that performs core hand motions—pinching, rotating, hooking, and grasping—through a simple design using a coiled spring, shell, and pneumatic system, rather than replicating human hand anatomy.
The device exceeded human performance in some metrics, extending and contracting 3.5 times more than a human hand and grasping objects up to almost 13 times larger than comparable systems, suggesting that simple designs can achieve high dexterity at low cost across industries including aerospace inspection and laboratory work.
What happened
Researchers at Chinese institutions published a study in Advanced Science describing the BioflexBot, a robot gripper that replicates core hand motions—pinching, rotating, hooking, and grasping—using only a coiled spring, constraining shell, and basic pneumatic system with two inputs, rather than mimicking human hand anatomy.
Why it matters
Most robotic hands attempt to replicate the complex structure of human hands, which drives up hardware and control costs; the BioflexBot achieves comparable dexterity by mimicking function instead of form, suggesting that simpler designs can deliver high dexterity at extremely low costs for applications across industries, potentially including aeroengine inspection, humanoid robot integration, and laboratory tasks.
What to watch
The BioflexBot extended and contracted 3.5 times more than the human hand and could grasp objects up to almost 13 times bigger than similar systems; researchers plan to translate the prototype to a fully automated platform in future work.
The study, published by Wiley in the journal Advanced Science, emerged from a research collaboration among Chinese universities seeking to address the limitations of traditional robotic hand design. Rather than replicating the complex skeletal, muscular, and neurological structures of the human hand, the team—including senior authors Yingtian Li of the Chinese University of Hong Kong, Shenzhen and Yang Yang of Nanjing University of Information Science and Technology—engineered the BioflexBot to achieve the same functional outcomes using a dramatically simpler architecture: a coiled spring, a constraining shell, and a basic pneumatic system (which uses compressed air to drive mechanical action). Crucially, this design required only two pneumatic inputs to enable four core hand motions: pinching, rotating, hooking, and grasping.
The researchers validated BioflexBot through a series of carefully chosen tasks that span laboratory, household, and industrial contexts. For pinching validation, the robot successfully manipulated an acupuncture needle and reliably transported liquid using a pipette—delicate operations typical in healthcare and laboratory settings. To test rotation, BioflexBot rotated a bottle cap, achieving almost four times the rotation range of a human hand. The team demonstrated hooking capability by having the robot engage objects such as a toolbox and goggles. For grasping, the BioflexBot securely gripped objects of varying sizes, up to almost 13 times bigger than comparable robotic systems could handle.
Beyond matching human hand performance, the BioflexBot exceeded it in key metrics. The robot extended and contracted 3.5 times more than a human hand, enabling it to grasp complex objects, reach long distances, deliver objects in confined spaces, and transport multiple objects sequentially. The researchers demonstrated three concrete applications: inspecting aeroengine blades (using the robot's large-curvature distal bending to examine blade trailing edges), completing daily tasks when integrated with a humanoid robot body, and conducting chemistry experiments. According to the team, these findings indicate that the BioflexBot's simple design delivers high dexterity at extremely low costs, with potential applications across multiple industries. They announced plans to translate the prototype to a fully automated platform in future work.
The BioflexBot research addresses a longstanding challenge in robotics: replicating human hand dexterity without replicating human hand complexity. Traditional robotic hands attempt to mirror the intricate anatomy of biological hands, which has historically resulted in systems that are expensive to manufacture and difficult to control. By contrast, the BioflexBot team—led by researchers from the Chinese University of Hong Kong, Shenzhen and Nanjing University of Information Science and Technology—pursued a functional approach: they identified the core motions a hand must perform (pinching, rotating, hooking, and grasping) and engineered a system that achieves those motions through minimal mechanical components: a coiled spring, a constraining shell, and pneumatic actuation with just two control inputs.
The validation experiments demonstrate that this minimalist design does not sacrifice performance. The BioflexBot reliably handled delicate laboratory and healthcare tasks (moving acupuncture needles, transporting liquids via pipette) while also exceeding human capabilities in extension, contraction, and load capacity. The device rotated a bottle cap nearly four times farther than a human hand could, and grasped objects up to 13 times larger than those comparable robotic systems could handle. These results suggest that biomimicry in robotics need not be anatomically faithful to be functionally superior. The researchers frame their findings as evidence that high dexterity can be achieved at low cost—a proposition that could reshape how roboticists approach gripper design for humanoid robots, industrial automation, and specialized inspection tasks in confined or hazardous environments.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
Lockheed Martin is using machine learning to replace static flight-hour forecasting with dynamic demand predic…
Hugging Face published a walkthrough showing how to record robot demonstrations into a Storage Bucket, train a…
Foxconn announced on August 12 at its second-quarter 2026 earnings conference that it is broadening its busine…
South Korea plans to establish a strategic investment account with at least 20 trillion won in assets within t…
Nvidia CEO Jensen Huang signed seven Japanese companies—Fujitsu, FANUC, Yaskawa Electric, Kawasaki Heavy Indus…
LTX released LTX-2.5, an open-weights world model designed to generate video and simulate how physical environ…
The AI news that matters, in one minute each morning.
Sign up free