Squishy robots can now heal themselves and grow as they explore

Credit: Thomas Fuchs

Constructed from delicate, flexible and lifelike materials, soft robots have the potential to improve on their clunky, metal-bodied predecessors. Such machines could more nimbly explore other planets, gently collect organisms from the ocean depths and even lend surgeons a hand. But stubborn design challenges have long held them back from making it out of the lab and into our lives. Now a new generation of soft robots is navigating, growing and self-repairing its way to meeting researchers’ lofty expectations.

Squishy materials let robots deform to adapt to changing environments, such as constricting tunnels. Soft robots can also handle fragile materials, such as human organs or brittle rocks, without crushing them. Even some mostly rigid-bodied robots, including the famously agile walkers made by Boston Dynamics, incorporate soft parts for better movement. Many developments in soft robotics are inspired by traits of living organisms, such as octopuses’ flexibility or the high water content of jellyfish. And new designs seek something less tangible: animal-like independence.

“The robotics community has been continuously focusing on the science and engineering of autonomy,” says Massachusetts Institute of Technology roboticist and computer scientist Daniela Rus. “We have made advancements on the soft body components and also on the algorithmic control … and we are now using these advancements to make increasingly more capable and self-contained autonomous soft robots.”

When independently exploring treacherous territory, soft robots are more prone to cuts and punctures than rigid machines are. One group of researchers, inspired by the self-healing properties of human skin, recently created an experimental robot that can bounce back from small injuries. The team described its study findings in Science Advances.

“If we have our druthers and achieve robots that operate for years at a time while performing dexterous tasks, then many opportunities open up for us,” says study co-author Robert Shepherd, an engineer at Cornell University. “One clear example is space exploration—perhaps building research habitats on the moon or even surveying the oceans of Europa. In these remote operating environments, robots will accumulate damage and may not have anyone around to repair them.”

Shepherd and his team designed a soft robot that not only heals damage but doesn’t need to be told when to do so. Using fiber-optic sensors, the robot can detect when its material has been punctured. Then it uses a hyperelastic material, called polyurethane urea elastomer, to quickly heal the wound. The robot is also programmed to move in a new direction after damage—ideally escaping whatever caused it. Later work could expand these repairs to bigger missing chunks and holes.

Another team created a soft robot that “grows” like a plant or fungus for a study published last year in the Proceedings of the National Academy of Sciences USA. Growing robots could burrow underground or lay new infrastructure on other planets. But to grow, soft robots typically have to drag material behind them and use it to 3-D-print new structures. This can hinder a robot’s work like lugging around a garden hose would for a person, says study co-author Chris Ellison, a University of Minnesota engineer and materials scientist. “If you drag your garden hose, and you turn a corner around a tree, the force on the hose goes up,” he says. And it continues to increase exponentially with each bend.

The researchers turned to plants for a solution. “They don’t extend their roots by dragging more roots behind them,” Ellison says. “They transport liquids, and then they transform those liquids to solids, and that ultimately is what builds a structure.” His team’s new robot uses light to solidify a liquid while spitting it out of a small hole to form a tube, which extends from its launching point to wherever it needs to go. The robot can control the tube’s shape as it grows, allowing for navigation of complicated paths without running into the garden hose problem. Robots might one day use this technology to smoothly inspect pipes underground or to pass through the human body for medical applications, Ellison adds.

Engineers have also made major progress in improving soft robots’ sensing and motion abilities, which will aid deployment in remote environments. For example, Rus’s group recently built a robot with networks of air-filled channels throughout its body. It can measure pressure changes within these channels to determine where its body parts are in space, similar to human proprioception. Other groups have experimented with various types of sensors, artificial muscles and machine learning to create smoother movement and precise perception.

Building soft robots that can work, heal and grow independently could change many areas of human life. “Soft robot hands are enabling a new age for manufacturing,” Rus says. Dexterous robots could fit into factory settings more easily if they had humanlike hands that could use the same tools we do, notes ETH Zürich roboticist Robert Katzschmann, who was not involved in the above studies.

Soft robots could also find a place in hospitals. Working alongside nurses and doctors, a robot could help softly and safely hold organs in place during surgery. “Helping hands could make medicine a bit less costly,” Katzschmann says, “so you don’t need 10 people in an OR. You could do with just one or two.” Ellison’s team says its robot could someday grow through tissue and search for cancerous tumors, potentially replacing a dangerous surgery altogether.

“I think soft robots are an avenue to endurance and agility not seen before in artificial machines,” Shepherd says. With heightened sensing and motion skills, robust compositions, and newfound independence, these squishy machines’ future looks solid.

This article was originally published with the title “Soft Power” in Scientific American 328, 3, 12-13 (March 2023)

doi:10.1038/scientificamerican0323-12

ABOUT THE AUTHOR(S)

    Nora Bradford is a freelance science writer and a Ph.D. student in cognitive science. Follow Bradford on Twitter @noraabradford

    Read More

    Leave a Reply

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