Researchers create most human-like robot skin yet

Researchers create most human-like robot skin yet

Bipedal, humanoid robots from companies like Boston Dynamics seemingly advance by the week, but they still can’t match their most audacious sci-fi inspirations. One of the major barriers they haven’t overcome is the ability to “feel” sensations like a human. Although researchers have tried various sensors to give robots a rudimentary sense of touch, these systems are often costly, inaccurate, and limited to detecting only one type of sensation at a time.

But that may be about to change. Researchers from the University of Cambridge and University College London have developed a new type of responsive “synthetic skin.” The skin is made from a single hydrogel that is capable of detecting touch, pressure, heat, cold, and damage in a way that more closely resembles human skin. 

The researchers were able to cast this gel into the mold of a hand and place it over a robot like a glove. By wearing the chunky flesh mitt, the robot can more accurately “feel” the world around it. Though still in the early stages, researchers say this type of synthetic skin could one day help robots work more effectively in factories, dangerous environments, or even assist in disaster relief efforts. The findings were published this week in the journal Science Robotics

Synthetic skin in robotics isn’t entirely new. Since as early as 2016, researchers have demonstrated the ability to use small sensors in robotic fingertips and hands to help them sense the shape and texture of certain objects. Until now though, that process would get quite complicated, quite quickly. A robotic hand capable of sensing both heat and pressure would require two different types of sensors—each designed to detect a specific sensation. These sensors can interfere with one another, causing what researchers call “cross talk.” Such mixed signals can also reduce the accuracy of perception and, in some cases, even damage the sensors. Adding more sensors to a small area like a robot hand also increases complexity, which raises the likelihood of errors.

“Having different sensors for different types of touch leads to materials that are complex to make,” lead author and Cambridge professor David Hardman said in a statement. “We wanted to develop a solution that can detect multiple types of touch at once, but in a single material.”

The researchers aimed to develop an alternative to the complicated multi-sensor approach by using a single material capable of detecting multiple types of stimuli. To do that, they used a soft, stretchable, and conductive gelatin-based hydrogel that can convert physical inputs, like heat and pressure, into electronic signals processed by a computer. According to the researchers, the gel itself is capable of detecting signals through 860,000 individual pathways embedded within the material. They conducted physical tests on the synthetic skin, exposing it to various types of stimuli at different levels to fine-tune its responsiveness. 

Related:[Lab-grown, self-healing human skin designed to cover robot faces]

chart of how hand works
Image: University of Cambridge

That synthetic skin can also be melted down and reconfigured into new shapes. In this case, the researchers wanted to see if it would still function when molded into the shape of a human hand. The result was an odd-looking yellow-and-black hand resembling a battered construction glove. This glove was then placed over the robot’s hand. To test it, researchers pressed on it with a finger and lightly swabbed the surface. Even in this more realistic form, the skin was still able to differentiate between different levels of pressure.

From there, the researchers exposed the hand to a “heat blast,” partially melting it. They took the abuse a step further by using a scalpel to slice the hand open. Thanks to the synthetic skin, the glove was able to “feel” each different stimulus—all through a single, universal sensor. While it was still less responsive than human skin, it outperformed other approaches that rely on multiple sensors.

“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,” paper co-author and University College London researcher Thomas George Thuruthel said in a statement. “Our method is flexible and easier to build than traditional sensors, and we’re able to calibrate it using human touch for a range of tasks.”

robot hand that looks burned
Researchers poked, burned, and slashed a robot’s hand to see if it could feel. Image: University of Cambridge Image: University of Cambridge

A path towards robots ‘feeling’ their way around the world 

The researchers say there’s a practical reason for all the poking, prodding, and burning their robot’s weathered hand. The tests demonstrate that, at least in theory, this synthetic skin could be melted down and molded to fit various parts of a robot’s body and still sense touch. A machine with this outer layer of sensing skin could then navigate its environment—whether in an automobile factory or on a construction site—and interact with the world in a way more similar to a human. That capability could become especially useful if humanoid robots, like those being developed by companies such as Figure and Tesla, are ever deployed to work alongside humans in real-world settings. If a robot is handing a car part to a human, it should probably know if it’s ripping hot.

Figure is already testing its humanoid robots in BMW’s South Carolina manufacturing facility. Amazon, meanwhile, is reportedly already training humanoid robots to carry and drop off packages to customers. 

An odd future where fleshy robots deliver your latest order of toilet paper, in other words, might not be all that far-fetched. 

 

More deals, reviews, and buying guides

 

Mack DeGeurin is a tech reporter who’s spent years investigating where technology and politics collide. His work has previously appeared in Gizmodo, Insider, New York Magazine, and Vice.


Leave a Comment

Your email address will not be published. Required fields are marked *