Small robots typically struggle on soft ground. Their motors may be strong enough to move their legs, although each step can lose force as the feet slip or sink. A new frog-like robot gets around the problem by winding up two slender metal rods and snapping them against the ground, launching itself into the air with each hop.
In recent testing, the 4.3-inch-long robot hopped across wood, cloth, acrylic, leather, grass, and sand. On those six surfaces, it traveled an average of 2.46 body lengths per second, roughly three times the speed of a comparable robot fitted with rigid legs. The difference was especially pronounced on cloth and grass, where the rigid-legged machine nearly stalled.
Those results, reported in a study published in Science Advances, point to a way of getting brief, forceful movements from small motors. The motors load energy into the rods over time. Then the rods release it in a fraction of a hop.
The trick is getting a rod to snap at all. A bent, twisted rod does not necessarily spring into a new shape when pushed farther. Depending on its geometry and how its ends are moved, it may simply buckle gradually. That difference matters to a robot. Gradual bending spreads the motion out, while a sudden snap delivers a concentrated shove.
Researchers at UCLA and the University of Michigan investigated how to choose between those outcomes. They combined mathematical models, computer simulations, and automated experiments in which a robotic arm deformed elastic rods. From those tests, they mapped the combinations of shape and loading that produced a sharp, repeatable snap.
The robot’s limbs put that map to work, each of which contains a thin rod of Nitinol, a flexible nickel-titanium alloy, held in a helical shape. A small motor rotates one end of the rod. As it bends and twists, the rod stores elastic energy until it reaches a tipping point and snaps into a new shape. That sudden motion pushes against the ground and launches the robot into a hop. While airborne, the limb resets for the next push.
“Once we could predict when a rod would snap, we could use that sudden release of energy to turn a simple motor movement into a powerful push that sends the robot hopping forward,” UCLA graduate student and study co-author, Dezhong Tong, said in a press release.
The prototype weighs 98.2 grams, about 3.5 ounces, and carries its own battery and two motors. It completed about two hopping cycles per second. On wood, its fastest measured surface, it reached 3.21 body lengths per second. For a robot 11 centimeters long, that works out to about 14 inches per second. Its hops rose to a little more than its own body height.
The comparison with rigid legs offered the clearest test of what the snapping rods contribute. The researchers kept the body, battery, motors, foot-contact material, and motor commands closely matched, changing the elastic limbs to rigid printed ones.
On grass, the snapping robot traveled at 1.91 body lengths per second; the rigid-legged version managed 0.22. On cloth, the speeds were 2.50 and 0.10 body lengths per second, respectively.
Significantly, the improvement did not come from giving the hopping robot a much larger average electrical input. Under matched commands, it drew 1.63 watts, compared with 1.41 watts for the rigid-legged version. Instead, the elastic limbs accumulated energy and released it in a stronger burst. Force measurements found a higher, sharper upward push from the snapping robot.
That difference helps explain its performance on soft surfaces. A rigid leg depends more on continued contact with the ground. The snapping limb delivers a quick kick, then lifts off. It still needs enough traction to launch, but it spends less of each stride pushing against material that can give way beneath it.
Researchers also took the robot beyond the six-surface speed tests. The tiny machine could hop up and down a constructed set of steps and follow a teleoperated route through outdoor sand scattered with rocks.
With its electronics sealed and flexible fins attached, the robot also swam at about half a body length per second. Controlling its two limbs separately let it turn, while timing changes produced repeated backflips. A simple light sensor demonstration showed it could steer toward a light source.
These are prototype demonstrations, instead of evidence that the robot can independently pilot a real-world mission. Its swimming setup required modifications, and researchers controlled its movement remotely through the rocky sand.
The researchers also note that each snap has to be reset and timed with the robot’s landing and takeoff. A limb built for a powerful burst cannot vary its output as smoothly as a conventional motor-driven joint.
Still, the findings deliver more than just the novelty of a tiny jumping robot. Its central result is a way to predict when an ordinary elastic rod will bend quietly and when it will snap. That gives designers a way to choose the burst before they build the machine around it.
Ultimately, for a tiny robot facing grass, loose sand, or a step taller than its body, the shape of a thin metal leg may matter as much as the motor turning it.
“Because it’s the [sic] rod’s shape — not its size — that determines whether it snaps sharply or deforms gradually, the same design rules apply across a wide range of scales,” co-author and UCLA professor of mechanical and aerospace engineering, Dr. Khalid Jawed, explains. “This opens a promising path toward robots just a few millimeters wide, turning small motor movements into powerful bursts of motion.”
The study, “Geometry-controlled instability pathway selection in elastic helices enables fast, efficient robotic locomotion,” appeared in Science Advances.
Tim McMillan is a retired law enforcement executive, investigative reporter and co-founder of The Debrief. His writing typically focuses on defense, national security, the Intelligence Community and topics related to psychology. You can follow Tim on Twitter: @LtTimMcMillan. Tim can be reached by email: tim@thedebrief.org or through encrypted email: LtTimMcMillan@protonmail.com