Soft Robots for Hard Problems

27 March 2012—The word robot calls to mind images of C-3PO or the Terminator. But robots don’t necessarily need to be gleamingly metallic and hard-edged; some might even be downright squishy. That at least is the vision of some robotics researchers, including Carmel Majidi, assistant professor of mechanical engineering at Carnegie Mellon University, in Pittsburgh, and head of the school’s Soft Machines Lab.

“Nature is just full of examples of functionality without any rigid parts,” Majidi said during the American Physical Society’s March meeting in Boston. Think of an octopus squeezing through a tight opening or a Venus flytrap snapping shut on an unwitting insect.

Soft robots could be built from various types of rubber or silicone. Adam Stokes, a postdoctoral fellow in George Whitesides’s lab at Harvard University, showed off a soft rubber gripper picking up an uncooked egg and an anesthetized mouse without damaging them. The gripper, a small six-pointed star, is made of two types of rubber, one softer and more extensible than the other. Air is pumped into microchannels in the device through a small tube. “When you put air in, the softer rubber extends more and introduces curvature,” Stokes explains. The star bends around the top of the egg, tight enough to lift it but with enough give so the egg doesn’t break.
Majidi says the field of soft robotics is still fairly new and that researchers need to find alternatives to air pumps as a way to control the devices. The robots will also need ways to sense their own position. For that he’s exploring the use of microfluidics, specifically liquid-filled microchannels inside a film of rubber. Something as simple as saltwater would render the channel conductive so that the device would become electronic. But there are other fluids that could work, such as Galinstan—an alloy of gallium, indium, and tin that’s liquid at room temperature and a million times as conductive as saline, making it comparable to copper wire.

Because bending or stretching such a circuit changes the shape of the microfluidic channel, it also changes the circuit’s conductivity and thus alters an electrical signal passing through it. “You get something that functions like a stretchable circuit,” Majidi says. Such a device could act as a sensor that measures strain, pressure, or curvature. It could even be used as a stretchable antenna, as was recently demonstrated by a team led by Michael Dickey at North Carolina State University.

And soft devices would be comfortable for human use. “You can wear them, essentially,” Majidi says. He envisions a sensor worn over a finger or other joint to monitor body motion. Or the soft circuits could be integrated with textiles to create a wearable keyboard or other “smart” clothing.

“There are lots of things you could do with this,” says Stokes. Perhaps a robot could be a hybrid of hard parts, performing some of the functions robots perform now, plus soft parts for manipulating delicate objects.

His group also makes a four-limbed, air-powered robot that can crawl along a benchtop or bend itself to fit through a narrow opening. Stokes says they’re studying the role that different gaits could play in letting robots move at different speeds.

Soft-robot parts are relatively easy to build, the researchers say. The Harvard lab, for instance, uses a 3-D printer to make a mold. Stokes says that by selecting materials with the right mechanical properties, it should be possible to make devices in a wide range of sizes.

The ways in which soft robots differ from their hard-bodied forebears—from their pliability to the fact that they move very differently from the traditional machines based on joints and bearings—give robotics researchers entirely new avenues to explore, many inspired by nature, Majidi says. “We can start building devices that are more multifunctional, more versatile,” he says.