The takeaway: Researchers in China have developed an artificial skin that allows robots to sense pressure and respond in ways similar to living tissue, helping them avoid damage. The flexible "neuromorphic robotic e-skin," or NRE-skin, applies the same signaling principles used by human nerves to artificial materials, advancing tactile sensing for both robots and prosthetic devices.
Human skin transmits sensory information as electrical pulses, or spikes, that encode signals related to pressure and pain. NRE-skin mimics this biological process by converting pressure inputs into spiking signals that can be processed in real time.
The system consists of a stretchable polymer base embedded with pressure sensors, which are connected via conductive polymers to a circuit that translates mechanical force into voltage spikes. These spikes carry information across four characteristics: shape, magnitude, duration, and frequency. As in biological neurons, pressure intensity is primarily encoded through spike frequency, while the remaining features act as unique identifiers for each sensor.
Each sensor also emits periodic "I'm still here" pulses to confirm proper operation. If these signals cease, the network flags the corresponding area as damaged or disconnected.
Once pressure data reaches the processing unit, the system determines both the force intensity and its location. It also evaluates whether the input exceeds a predefined pain threshold. When that threshold is crossed, the system generates a pain signal that triggers reflex-like movement without involving a central controller.
In testing, a robotic arm covered with the e-skin recoiled in response to excessive pressure, while a connected robotic face altered its expression based on the detected force.
The next processing layer filters incoming sensory signals before forwarding them to the central control unit, where AI software governs more complex behaviors. Researchers calibrated the pain threshold by comparing sensor input to pressure levels that humans typically perceive as painful, allowing the system to simulate a realistic damage response.
Each e-skin patch functions as a modular tile that connects magnetically to adjacent segments. These magnetic contacts automatically align wiring, link data channels, and register each segment's identity code. If a patch fails, it can be easily replaced and re-registered within the robot's control system.
Despite being described as "neuromorphic," the e-skin does not replicate the full architecture of biological nerves. Instead, it draws inspiration from how neurons communicate, rather than attempting to duplicate their physical structure. At present, the material senses only pressure, whereas human skin can distinguish among multiple stimuli, including heat, cold, and chemical irritants. Adding such capabilities would require additional processing layers to keep different signal types separate.
Even so, NRE-skin integrates naturally with spiking neuromorphic processors – chips designed to run neural networks using low-power, event-driven electrical impulses. Pairing these processors with the new skin could enable energy-efficient robots capable of real-time tactile responses.
Overall, the approach represents another step toward giving machines something that behaves – if not truly feels – like a sense of touch.