Elizabeth Watson Illustration

A team of researchers at the Yale School of Engineering and Applied Sciences successfully improved the dexterity of soft robots using a method called tensile jamming.

Soft robots, robots made from pliable materials, are typically limited to a specific range of motions determined by their original design. The team worked to create soft robots capable of bending in different directions, marking an improvement on past technologies. With a bundle of soft fibers and a vacuum pump, the researchers were able to control the stiffness of the actuator — a device responsible for controlling motion — and created soft robots dextrous enough to handle a Rubik’s cube and twist the lid off of a jar.

“If you can selectively control where and how much stretching occurs across the surface of your robot, then you can control how it moves when it’s interacting with things and when you’re controlling it with your actuators,” Dylan Shah GRD ’23, one of the study’s co-authors, said. 

According to lead author of the study, Billy Yang GRD ’25, many traditional soft robots function by pumping fluid into a single chamber, causing the chamber to bend. The materials surrounding the chamber determine the direction of this bend, as the chamber will bend towards whichever wall is less stiff. Unfortunately, this design limits soft robots to one set of movements, which are determined by the stiffness of its original materials.

In order to overcome this limitation, the team decided to surround the actuator with a bundle of quickly-stiffening tensile fibers. The fibers were partially made of stretchable silicon and non-stretchable polyester. When the entire system is under a neutral, or atmospheric pressure, the materials are easily stretched. But when the system is “jammed,” or placed under pressure by a vacuum, the fibers press together and the polyester sides link together. Because the motion of soft robots is dependent on the stiffness of its materials, changing which fibers are jammed and which fibers are not jammed will change how the robot moves.

“[The fibers] sort of bunch together and then provide a bunch of stiffness, whereas at atmospheric pressure without the vacuum you can sort of stretch them freely,” explained Robert Baines GRD ’23, the other lead author of the study.

According to Yang, the tensile stiffness was about 20 times greater while pressure was being applied. By applying vacuum pressure to different groups of fibers, the researchers were able to change the stiffness in different parts of the actuator and increase the dexterity of a soft robot finger. According to Yang, this finger was able to bend in 36 different directions — the most that any soft robot finger has been able to achieve.

When three of these fingers were combined to create a gripping device, the robots were able to successfully grasp different objects, including handling a Rubik’s cube and twisting the lid of a jar.

Baines noted that the vacuum allows for this transition to occur very quickly, in less than one-tenth of a second, so this method could easily be applied to existing soft robots and other technologies that require rapid changes in stiffness. While this speed is impressive, a potential area for improvement is increasing stiffness of the fibers while they are under pressure.

The direction of future work will be to create tensile fibers and actuators that are bigger and smaller, according to Yang. He hopes that smaller versions of this system could potentially be used for minimally invasive surgery, while larger versions could have applications for soft exoskeletons.

According to Yang, this study is part of a four-year grant from the National Science Foundation, or NSF, that was awarded to engineering professor Rebecca Kramer-Bottiglio. She was awarded the grant as part of the NSF’s Emerging Frontiers in Research and Innovation program, with the funds intended to support the development of shape-changing robots. The goal of this study, like many within this project, was to create robots with similar capabilities to biological systems.

“We as roboticists or engineers are always trying to mimic nature, in terms of nature’s ability,” Yang said. “I think this work puts us … a step closer to doing that, by making soft systems that are able to be as dexterous as some of these organisms or structures that we see in nature.”

The results of this study were published in the journal Science Advances on Oct. 1.