Engineers at Queen Mary University of London have developed a stretchable material that changes color under pressure, giving robots a high resolution sense of touch that can be read with nothing more than an ordinary USB camera. The university describes the work as a step toward solving one of the persistent gaps in robotics: machines that can see and hear but cannot feel how hard they are gripping.
How it works
AI systems have advanced rapidly in language, vision and decision making, but touch has lagged. Most robots can see an object yet struggle to judge how firmly they are holding it, whether it is soft or hard, or whether they are about to damage something fragile.
The Queen Mary team’s answer is a mechanochromic material, one whose color shifts in response to mechanical force. When the material is pressed, the pattern of colors reveals where pressure is applied and how much. A standard camera watches the surface and translates the color changes into a pressure map in real time. No embedded electronic sensors, wiring or heavy computing are required.
In practice, a robot hand covered with the material would display different colors as it grips an object. The camera reads those colors and tells the robot how tightly it is holding, where pressure is concentrated and where it is light, approximating the feedback a person gets from their fingertips.
Fingerprint level detail
The researchers report a spatial resolution of roughly 100 micrometers, fine enough that the ridges of a human fingerprint are clearly visible in the sensor’s output. That is a considerable improvement over many existing robotic touch sensors, according to the university.
Potential applications
The team points to several uses. Factory robots could handle delicate electronic components more safely. Prosthetic hands could give wearers a richer sense of contact. In robotic surgery, instruments could detect subtle differences between tissues rather than relying on vision alone, potentially making procedures safer and more precise.
The work also reflects a shift in the wider field. After years in which the largest AI companies concentrated on software intelligence, attention is moving toward embodied AI: robots that perceive and act in the physical world, including through touch. Low cost, high resolution sensing of the kind demonstrated here is one of the building blocks that direction requires.
