In 2017, a humanoid robot developed by University of Tokyo researchers dropped down and did some push-ups.

The 115-lb technology, named “Kengoro,” even appeared to perspire, circulating water through its porous aluminum frame.

By “sweating," Kengoro did not require the clunky components typically installed to cool down a robot's motors, like fans or heat sinks.

Soft robots, though flexible, retain more heat than their metal counterparts — and require more sophisticated ways of preventing overheating.

Researchers from Cornell University have been working out an even better way to get a sweat going.

The robotics team, led by associate professor of mechanical and aerospace engineering Rob Shepherd, manufactured soft robotic actuators with reversible pores. The actuators only sweat when they need to.

The team’s paper, “Autonomic Perspiration in 3D Printed Hydrogel Actuators ,” was published on Jan. 29 in the journal Science Robotics. The research was supported in part by the Office of Naval Research Young Investigator Program.

Shepherd’s team partnered with the lab of engineering professor Emmanuel Giannelis to create the necessary nanopolymer materials for the sweating soft-robot muscle. The 3D-printing technique, known as multi-material stereolithography, used light to cure resin into predesigned shapes.

The resulting finger-like actuators, made from two hydrogel materials, are spongelike and even somewhat "smart," both retaining water and responding to temperature.

How smart? When temperatures reach above 30 °C (86 °F), the material's poly-N-isopropylacrylamide base layer shrinks. The reaction pushes water up into a top layer of polyacrylamide that is perforated with micron-sized pores. The pores, sensitive to the same temperature range, automatically dilate to release the “sweat,” then close on their own when the temperature drops below 30 °C.

“The best part of this synthetic strategy is that the thermal regulatory performance is based in the material itself,” said co-lead author T.J. Wallin , M.S. ’16, Ph.D. ’18, and a research scientist at Facebook Reality Labs. “We did not need to have sensors or other components to control the sweating rate.”

Releasing the water reduces the actuator’s surface temperature by 21 °C within 30 seconds, a cooling process that is approximately three times more efficient than a human's, said the researchers.

The team incorporated the actuator fingers into a robot hand that could grab and lift objects. The autonomous sweating cooled not only the hand, but the object being held as well.

In this image, a three-fingered sweating gripper manipulates multiple hot objects. The soft gripper is capable of lifting a hot beverage can, perspiring, and cooling the beverage can down quickly. The same performance is shown by the gripper conformally grasping the foam brain (right) and cooling it down by sweating. ("We found the cooling rate 70C to 50C was approximately 2x faster when the object was manipulated by our sweating grippers," said Wallin.)

In an edited interview with Tech Briefs below, Wallin explains why this kind of thermal regulation demonstrates such a promising future for soft robots and soft robotic actuators.

Tech Briefs: When we say the robot “sweats,” what is actually happening?

T.J. Wallin: This is an example of a fluidic elastomer actuator. Basically, the actuator is a polymeric water balloon that bends in response to pressurization with water. The actuator design is such that there is an internal channel that we can pump in water to fill and increase the volume of the actuator.

The multi-material construction enables regions of different stiffness to direct the actuation. To make the actuator sweat, we used both novel materials and advanced manufacturing. The top layer of the actuator contains tiny micropores made out of a “smart gel” which reversibly opens and closes in response to heat. When the material is hot, the pores are open and the internal water leaks out to cool the actuator.

Tech Briefs: How is your method different from the Kengoro robot?

T.J. Wallin: The researchers have already made a similar sweating robot, Kengoro,  out of rigid materials. This robot leaked water through its aluminum skeleton and, because of this cooling, was able to do longer bouts of activity without motor failure. We are different in the sense that our sweating strategy is compatible with soft materials and soft robots, which can be more akin to biology.

Tech Briefs: What have been the challenges that have prevented this kind of “sweat” method from being used in a mainstream way?

T.J. Wallin: There are numerous benefits to working with soft robots, like more sophisticated function with simpler design and fabrication, as well as safe interfacing with biology. In the context of soft robots, thermoregulation hasn’t really been tackled yet. That is because the conventional devices that dissipate heat (heat sinks, fans, radiators) are built from rigid materials that are incompatible with soft robots.

Additionally our design is an example of embodied intelligence. Unlike the leaky skeleton from the robot Kengoro, our pores reversibly open and close due to a material response to changing temperatures. Our actuators only sweat when they need to cool.

Tech Briefs: What gets cooled down exactly? And is it possible for the water to do any damage to a robot? What prevents any damage?

T.J. Wallin: Our robotic hand can cool down both the robot and what it is manipulating. Water does not damage these soft robots because their materials and actuation mechanism are both robust to water.

Tech Briefs: Why is this method more valuable than traditional cooling methods like fans?

T.J. Wallin: This method is valuable to soft robots for a variety of reasons. Soft robots are constructed from polymeric materials which are thermally insulating. If we envison a future where these soft robotic bodies contain high torque motors or other devices that need to operate at peak power for long periods of time, then heat transfer will become an issue. All materials display some degree of temperature dependence. For example, crosslinked rubbers tend to get stiffer as temperatures rise, which would change how the robotic body responds to loads. To ensure we maintain stable operation and access the benefits of soft robots, we needed to find a strategy that could be compatible with soft materials. Additionally, fans and heat sinks take up valuable space and add load which incur other costs to robotic endurance.

Tech Briefs: What is most exciting to you about this work with soft robotic actuators?

T.J. Wallin: One of the most surprising things was that when we compared our actuator’s thermoregulatory performance to animal systems we found a favorable comparison. When we took the normalized (by mass) evaporative water loss, we found we were more effective at cooling than humans or horses.

Tech Briefs: In what kinds of applications has this been tested out?

T.J. Wallin: This was fundamental research without a specific application in mind. What is unique about our work is that we recognized the need to create thermoregulatory strategies compatible with soft robots as the current cooling mechanisms (heat sinks, fans, radiators) are made from rigid devices. We know that as soft robots continue to develop thermoregulation will become a more important component to enabling high-power, long-duration applications.

Other contributors included postdoctoral associate and co-lead author Anand Mishra; postdoctoral associate Wenyang Pan; doctoral student Patricia Xu; and Barbara Mazzolai of the Italian Institute of Technology’s Center for Micro-BioRobotics.