The Dawn of Bendy, Squishy Robots

At the March Meeting, the metal contraptions of the popular imagination made way for soft robotics powered by fluids and formed using origami folds.

By Sophia Chen | May 11, 2023

Credit: BYU

A researcher studies a prototype of a solar panel array that folds up in the style of origami, a paper craft of choice for some researchers in soft robotics. This prototype was designed by researchers at NASA and Brigham Young University in 2014.

---

The robot, according to emojis and The Jetsons, is a shiny, angular-jointed, metal machine. It contains a computer chip that runs software to control its motion, and you can probably destroy it by pouring water on it.

But researchers are moving toward new paradigms for autonomous machines. Instead of making robots out of metal, they want to make robots out of soft, bendy materials like silicone, vinyl, and nylon fabric, to work more safely alongside humans. Some prototype machines can execute logic without any electronics, and they can still work after being submerged in water or run over by a truck.

Known as soft robotics, the field draws from many disciplines, such as materials science, biology, computer science, and even traditional crafts. Physicists have room to play around, from developing design principles to building the machines. At this year’s March Meeting, researchers presented advances in the field in a session called Morphing Matter.

James McInerney, a postdoc at the University of Michigan, presented research on design principles for origami, the traditional Japanese paper-folding craft. McInerney traces contemporary scientific studies of origami to around the 1980s, when engineers realized they could use origami techniques to efficiently load objects to launch into space. Both NASA and Japan’s space agency, for example, have used the so-called Miura fold to pack a solar panel array inside a rocket.

McInerney began studying origami as a graduate student, after discovering that it presented interesting unsolved problems in geometry. “Using origami, you change the form of a material,” McInerney says. “Once you change the form, you change the function. The stiffness changes; the amount it bends in and out of the plane changes.”

For example, researchers working with origami seek to control a property known as Poisson’s ratio. A Poisson’s ratio describes how a material behaves when you compress it along one axis. “Think of a banana,” says McInerney. “If you squeeze it, it’s going to pop in the other direction. That’s a positive Poisson’s ratio.” Most materials are banana-like in this way, so researchers are interested in materials with a negative Poisson’s ratio, meaning that when you squeeze it, it compresses along the opposite axis. The Miura fold, for example, imparts a negative Poisson’s ratio to a sheet. When you compress it in one direction, the whole object collapses into a smaller volume.

Credit: Rob Felt/Georgia Institute of Technology (Eureka Alert)

Many origami folding patterns, like the Miura fold, raise interesting questions for physicists.

---

McInerney is working to classify origami folds into “families” of crease patterns that share similar properties. He uses the periodic table of the elements as an analogy. He aims to show that certain geometrical patterns in origami result in similar material properties, just as elements with the same number of valence electrons share chemical properties. He is working with an experimentalist to validate his theoretical findings.

But when it comes to the craft itself, McInerney is less experienced. “In practice, I'm not a very good traditional origamist,” he says.

In another talk, Anne Meeussen, a postdoc at Harvard University, presented her efforts to build robots powered by fluids instead of electronics. In a fluid-driven robot, for example, tubes might fill with fluid to reach some threshold pressure, spurring valves to open and the robot to move.

From a theoretical standpoint, fluids behave like electrons do in electronics. In fluidic robots, the flow of a liquid or gas stands in for the flow of electrons; pressure replaces voltage. “If you look at the equations that describe an electronic network versus a fluidic network, they look the same,” says Meeussen, who has a background in theoretical physics. “The difference comes in when you start talking about applications.” It’s easier, for example, to design fluidic robots to be waterproof or robust to radiation, but electronics work much faster than fluid-driven machines.

Other researchers have demonstrated fluid-driven prototypes, including one group that used carbon dioxide gas to make a sweatshirt lift its own hood. But these efforts have tended to borrow design strategies from electronics, such as using sequentially and serially connected modular components to complete a task.

However, Meeussen’s team is experimenting with a different design strategy. Instead of piecing together the robot component by component, as in conventional electronics, they start by simulating a random network of fluidic tubes. Then, they task a machine learning-inspired algorithm to remove tubes until the network achieves the intended goal, such as having the network move an object. The process is more akin to sculpting, as it involves removing unnecessary parts from the system, rather than building it up like Legos.

Meeussen presented a simulated fluidic network, designed using this strategy, that “learned” to classify three species of irises. She encoded properties of each flower, like petal length, into fluid pressures, which she inputted to the network. Each species of iris corresponded to a specific output pressure. The network learned which tubes to remove to produce the output pressure corresponding to the correct iris species — and it classified the species correctly 96.7% of the time.

With these promising simulations, Meeussen is working with mechanical engineers and roboticists to make a real-world prototype of a fluid-driven robotic arm. “It's been amazing to work with people with such different backgrounds and see how we can complement each other,” she says.

Sophia Chen is a writer based in Columbus, Ohio.