Scientists Tortured New Robotic ‘Skin’ to See If It Felt Pain. It Did.

Scientists are trying to make robots feel pain. To get there, they’ve built a new kind of synthetic skin—then burned it, stabbed it, and poked it senseless just to see what it could take.

Made from conductive gelatin and packed with over 860,000 signal pathways, this new “e-skin” can detect a range of sensations including, touch, temperature, and actual physical damage. Researchers say it mimics the mechanics of human skin more closely than anything currently on the market—and unlike other materials used in robotic sensing, it’s cheaper, easier to produce, and more durable. It was detailed in a June study published in Science Robotics.

“We’re not quite at the level where the robotic skin is as good as human skin, but we think it’s better than anything else out there at the moment,” Thomas George Thuruthel, a co-author and robotics researcher at University College London, told Live Science.

Robots Can ‘Feel Everything’ Now Thanks to This New Skin

Right now, most synthetic skins require multiple separate sensors to detect different sensations. That approach creates reliability issues, drives up cost, and makes it harder to scale. But this e-skin uses just one “multi-modal” sensor that can interpret several kinds of input at once—pressure, heat, cold, damage—without the need for multiple components.

To test it, scientists melted the hydrogel into the shape of a human hand and wired it up with different electrode patterns. Then they torched it with a heat gun, jabbed it with a robotic arm, and sliced it open with a scalpel. The hand didn’t flinch (obviously), but the sensors did: researchers collected more than 1.7 million data points to train an AI system to recognize what kind of sensation the hand was “feeling.”

That kind of high-resolution sensitivity could be game-changing for prosthetics, where accurate touch feedback helps users regain a more natural relationship with their environment. It could also level up humanoid robotics, allowing machines to adapt to tasks that involve fragile materials, human contact, or hazardous conditions.

“Our method is flexible and easier to build than traditional sensors,” Thuruthel said, “and we’re able to calibrate it using human touch for a range of tasks.”

It’s just sensors and software, but something about it feels uncomfortably human.