Robots are good at moving, but they are famously clumsy about feeling. A human hand knows instantly that a cup is slippery, hot, or starting to slip, because skin is packed with sensors wired straight to the brain. Give a machine that kind of touch and you face a hard problem: electronics are usually built flat and stiff, on circuit boards, while a body is curved, soft, and in constant motion.
A research group in China has demonstrated an approach that sounds almost like weaving. A team led by Zhu Meifang, an academician of the Chinese Academy of Sciences and a professor at Donghua University in Shanghai, together with researcher Wang Gang, has made fibers that are themselves complete circuits. Their results were published in the journal Nature Electronics under the title "Programmable spinning of integrated circuit fibres."
The Core Problem: One Device Is Easy, a Circuit Is Hard
Flexible fibers that do a single electronic job are not new. Researchers have made threads that conduct electricity, light up, or sense pressure. The difficulty comes when you need many of those devices working together as a circuit — the way a single light bulb is simple but a computer needs billions of parts connected in precise patterns.
Going from a fiber device to a fiber circuit requires solving three things at once: how the materials are arranged, how each device is structured, and how the devices are electrically connected. Doing all three while the fiber is being formed is the barrier that has kept fiber electronics from becoming real systems.
How 'Microfluidic Encoded Spinning' Works
The team's answer is to treat the design of a circuit as instructions for a spinning machine.
Microfluidics is the control of tiny, precisely managed flows of liquid. In the Donghua process, different liquids carrying different functions — materials that conduct electricity, insulate, emit light, or act as a semiconductor — are fed together as a fiber is drawn. By carefully designing these flows, the researchers convert a circuit layout into a spinning program.
The control works in two directions. Across the fiber's cross-section, they tune the liquid properties to lay down stable, distinct functional layers — conductive, insulating, light-emitting, or semiconducting. Along the fiber's length, they vary which liquids flow and when, so the layers are continuous or interrupted exactly as designed, forming the electrical connections between devices. The circuit is, in effect, written into the fiber by controlling fluids. As the thread forms, its material layers, device structures, and circuit interconnections are all created at the same time — so the fiber itself becomes the circuit, with no need to attach a separate board.
What They Actually Built
Using the platform, the team constructed several functional building blocks: modules for colored electroluminescence (light produced electrically), analog signal processing, digital logic operations, and contactless spatial sensing. They then connected different device types — transistors, resistors, and capacitors — according to a design, all within a single fiber, to realize circuit functions such as a tunable filter and an inverter. An inverter is a basic logic element that flips an electrical signal, one of the foundational operations behind computing.
That progression matters. The work moves beyond making one kind of functional module at a time and shows that different devices can be designed to work together inside one thread — evidence that fiber circuits can be engineered to order rather than discovered by accident.
The Robot Test
The researchers also showed that the fibers can do useful sensing in a realistic setup.
They placed spatial-sensing fibers made by this method on the inside of a robotic gripper's fingers. When a target object approaches or moves, it causes a change in electric-field coupling; the fiber reads that change, the system calculates the object's position, and it drives the arm to track the target automatically. In another demonstration, the team used fibers to form a flexible interactive element. These are early proofs that the threads can serve as the sensory surface the "electronic skin" idea promises — a covering that perceives its surroundings while bending with the machine.
Why This Fits Embodied Intelligence
The term that keeps appearing in the Chinese reporting is embodied intelligence, often called embodied artificial intelligence (AI) — the idea of AI placed inside a physical body that senses and acts on the real world rather than existing only as software.
For such systems, electronics must adapt to complex shapes and constant movement, which is exactly where soft, bendable fibers have an advantage over rigid boards. If functional circuits can be integrated directly into thread, then sensing and signal-processing elements could, in principle, be woven into a robot's outer layer the way fabric is woven into clothing — covering large, curved, moving surfaces with electronics that flex rather than break.
What to Keep in Mind
A few qualifications keep the claim honest.
This is peer-reviewed work in a reputable journal, which supports the core result, but the demonstrations are laboratory-scale. The circuits shown are relatively simple building blocks — filters, inverters, sensing and lighting modules — not dense computational systems, and a single fiber is a long way from the millions or billions of reliable, uniform devices that a true computing fabric would require. Turning spun fibers into a durable, washable, mass-manufactured electronic skin that survives real-world wear, and wiring the outputs of many fibers together, still poses major engineering challenges. Descriptions of robot "electronic skin" describe a promising route and early demonstrations, not a product already fitted to working robots. Scaling microfluidic spinning from fine research fibers to consistent industrial production is itself unproven at this stage.
What to Take Away
The genuine advance is one of integration. Earlier flexible electronics could put a single function into a thread; the Donghua team has shown a way to design and form interconnected devices — transistors, logic, light, and sensing — inside one continuous fiber as it is spun, and has used those fibers to let a robotic gripper sense and follow an object.
A robot wearing a full-body, touch-sensitive fabric is not here yet, and the gap from an elegant lab fiber to a rugged manufactured skin is real. But the Nature Electronics result removes a key obstacle on the path: it proves that a circuit, and not just a component, can be built into thread. For machines that must one day feel the world as directly as we do, that is a meaningful thread to pull.