Read our COVID-19 research and news.

By studying sidewinding in this rattlesnake, researchers made robots better dune climbers.

By studying sidewinding in this rattlesnake, researchers made robots better dune climbers.

Hamid Marvi/Carnegie Mellon University

Sidewinder robots slither like snakes

In 2011, archaeologists needed to find out whether parts of ancient Egyptian boats were hidden in dangerously unstable humanmade caves by the Red Sea. Howie Choset, a roboticist at Carnegie Mellon University (CMU) in Pittsburgh, Pennsylvania, thought he had the perfect solution. He had built robotic snakes, so he outfitted one with a camera and sent it slithering through the cave's narrow opening. But when the robot got inside, “we faced a challenge that we had not seen before,” Choset recalls. The robot couldn't make its way up the cave's sandy slope, and the exploration was a bust.

Now, Choset has found a way to climb that mountain. By teaming up with researchers studying how live snakes move, he and his colleagues have determined what it takes to make snake robots go uphill, even on slippery, sandy slopes. These reptiles, real and robotic, are sidewinders—they move forward not by slithering, but rather by wriggling their bodies perpendicular to the direction of travel in a undulating S-shaped wave.

The motion is peculiar and, supposedly, “if you look too long at [a sidewinder] you’ll go mad,” says Daniel Goldman, a physicist at the Georgia Institute of Technology in Atlanta. And that motion will enable a snake to go uphill. But to maintain a grip on an incline, the snake modifies its motion to keep more of its body touching the ground as it moves, he, Choset, and their colleagues report online today in Science.

Those insights could lead to robots that do better on rough terrain than robots with wheels, such as a Mars rover. “We don’t presently have machines that can climb steeply inclined fragile ground, and this work suggests new ways to build machines that can,” says Daniel Koditschek, an engineer at the University of Pennsylvania who was not involved with this work but who develops legged robots for desert explorations. “The opportunities and benefits for robotics here are significant.”

Goldman specializes in studying how animals manage to move through sand and other granular materials that tend to give way. Curious about snakes, he and Hamidreza Marvi, now at postdoctoral student at CMU, went to Zoo Atlanta, where they filmed and analyzed the movements of six sidewinder rattlesnakes climbing in an enclosure with an inclined floor covered with sand. They varied the angle of the floor with each test and also observed several vipers, snakes that are not sidewinders, try to ascend the incline.

The vipers slipped and failed to move up the slope, whereas the sidewinders had no problem, the researchers report. As it moves, a sidewinder sends a horizontal wave down its body. At the same time, it undulates up and down. As a result, the parts of the body on the ground push off while the airborne loops reach upslope, where they then make contact to push off. Marvi found that on flat ground, at any moment the rattlers have 25% of their body in contact with the ground. But the snakes tune their motion to the terrain. On a slope of 10°, 40% of the body remains in contact with the ground. That fraction increases to 45% on a 30° slope.

In making the adjustment, a snake has to balance two factors. Too much contact and the reptile can’t lift the other parts of the body high enough to reach up the slope. Too little contact, and the sand gives way under the snake’s weight. “Apparently, these animals have found a sweet spot,” Koditschek says.

Choset’s group programmed the robot to move in a similar manner, then altered the waves to change how much of the robot’s body was in contact with the sand at any one time. It, too, made it up through the sand.


And by tweaking the two waves, “we can make our snake robots do what the snake can’t do,” Choset says, such as turn on a dime. These attributes may lead to robots that can snake their way through rubble in disaster zones to find trapped people or that can inspect nuclear power plants.

“I’m smitten with this paper,” says Adam Summers, a comparative biomechanist at the University of Washington Friday Harbor Laboratories who was not involved with the work. “The biology is informed by the engineering and vice versa.”

Choset, too, is quite pleased. He doesn’t have the permits needed to go back to the Egyptian caves, but has now used the robot on another archaeological expedition. There, “we did much better,” he says. Things are looking up for robotic snakes.