Legged robots are impressive — some can even do a backflip — but the machines often stumble on natural surfaces like wood chips, rocks, and sand.

That's a problem, says third-year PHD student Emily Lathrop, who works in the Bioinspired Robotics and Design Lab at University of California, San Diego , because those surfaces compose most of the world around us.

Lathrop and her team created stronger, flexible robotic feet that may someday support applications like search-and-rescue and space exploration.

"These feet can enable robots to walk more efficiently and effectively over new types of natural terrains, such as rubble in disaster zones, the surface of other planets, or even something as close to home as forest underbrush," Lathrop told Tech Briefs.

Lathrop and the UCSD researchers added a surprising internal structure to put pep in the robot's step: Coffee.

The team filled the feet — flexible spheres made from a latex membrane — with coffee grounds, as well as root-like flexible fibers to stabilize the foot.

The fibers (shown in blue lines in the image below) are attached to the ankle of the foot and protrude into the coffee grounds in an egg-beater like shape.

"Think sandpaper in thin rope form," Lathrop told Tech Briefs.

a diagram of the UCSD flexible foot, with the sandpiper-like, stabilizing fibers shown in blue
A diagram of the flexible foot, with the sandpaper-like fibers shown in blue. (Image Credit: UCSD)

The robotic feet are able to transition between soft and stiff states, which helps the feet conform around terrain obstacles. The coffee, in a still state, provide flexibility and free movement. When a vacuum is applied, the grounds push together and become rigid.

Through a mechanism called "granular jamming," the coffee grounds move back and forth, behaving as both a solid and liquid. When the feet make contact with the ground, the contents firm up and provide solid footing. When transitioning between the steps, the ground unjam and loosen. The support structures help the flexible feet remain stiff while jammed.

By outfitting a hexapod robot (shown at the top of the article), researchers tested the flexible feet on flat ground, wood chips and pebbles. An on-board system generated negative pressure to jam the feet, and positive pressure to unjam them. To stiffen the foot, a vacuum pump removed air from between the coffee grounds.

The feet also can be passively jammed, when the weight of the robot pushes the air out from between the coffee grounds inside.

the soft robotic foot from UCSD conforming to the rocky surface it is stepping on
The soft robotic foot conforms to the surfaces on which it steps, allowing the robot to walk faster. (Image Credit: UCSD)

The UCSD team found that passive jamming feet perform best on flat ground, but active jamming feet operate more effectively on loose rocks. The feet also helped the robot’s legs grip the ground better, increasing its speed. The improvements were particularly significant when the robot walked up sloped, uneven terrain, according to the engineers' study.

The researchers determined that the foot reduced by 62 percent the depth of penetration in the sand on impact, and reduced by 98 percent the force required to pull the foot out when compared to a fully rigid foot.

In a short Q&A with Tech Briefs below, Lathrop takes us to the day of the test and reveals the inspiration behind the flexible design.

Tech Briefs: What does a “traditional” robotic foot look like, and how is yours different, technologically?

Emily Lathrop: Most traditional robot feet are made of hard components, whereas ours are able to transition between hard and soft states. Our foot is composed of a latex membrane filled with coffee grounds and thin abrasive fibers. In its normal state the foot is soft and the coffee grounds are free to move around inside the membrane, but when a vacuum is applied the grounds jam together and the foot becomes rigid. In this way we can alter the stiffness of our foot on the fly.

the parts of the hexapod robot from UCSD, including microcontroller, pumps, and the granular feet
The parts of the hexapod robot. (Image Credit: UCSD)

Tech Briefs: What inspired you to use coffee?

Emily Lathrop: Coffee grounds are ideal because you need some type of small particulate material in order to achieve granular jamming, which is the basis for how we can change the foot stiffness on the fly. The basic principle is that loosely packed coffee grounds flow like a liquid but lock together like a solid when air is removed. You can see this in action when you buy a new bag of vacuum packed coffee and it feels solid, much like a brick, but when you cut it open all of sudden you can squish the back and reshape it. Coffee is also a great choice because it is light and relatively inexpensive.

Tech Briefs: Take us through the tests.

Emily Lathrop: We first looked at the ability of the foot to conform around an obstacle and the amount of shear force a foot could withstand without slipping for several different types of internal structures to put inside the soft feet. For tests on a robot, the first test that we used was a speed test, where we measured the speed that different types of feet can achieve over rocks, wood chips, and flat ground.

The second was a duty cycle test, which measured how different types of feet influenced the energy that the motors had to expend to move the feet through a desired foot trajectory.

The third test was a drawbar pull test, which measured how much force the foot was able to pull behind it without losing traction on various terrain types.

Tech Briefs: What inspired this idea?

Emily Lathrop: We were inspired to incorporate stiffness changing feet into our robot by looking at how animals use the softness of their feet to create stable footholds with their environment. For the internal structures, we were inspired by plants that use their roots to stabilize hillsides and sand dunes, as well as how civil engineers use piles to stabilize the base of houses.

Tech Briefs: What’s next?

Emily Lathrop: We are excited to look at other types of terrains that these feet might excel at, and how we can selectively control stiffness to balance robot effectiveness and energy costs.

What do you think of the flexible robotic feet? Share your questions and comments below.