Dutch scientists built a brainless soft robot that runs on air

It runs like a stotting gazelle and swims like a dog.

Most robots rely on complex control systems, AI-powered or otherwise, that govern their movement. These centralized electronic brains need time to react to changes in their environment and produce movements that are often awkwardly, well, robotic.

It doesn’t have to be that way. A team of Dutch scientists at the FOM Institute for Molecular and Atomic Physics (AMOLF) in Amsterdam built a new kind of robot that can run, go over obstacles, and even swim, all driven only by the flow of air. And it does all that with no brain at all.

Sky-dancing physics

“I was in a lab, working on another project, and had to bend a tube to stop air from going through it. The tube started oscillating at very high frequency, making a very loud noise,” says Alberto Comoretto, a roboticist at AMOLF and lead author of the study. To see what was going on with the tube, Comoretto set up a high-speed camera and recorded the movement. He found that the movement resulted from the interplay between the air pressure inside the tube and the state of the tube itself.

When there was a kink in the tube, the increasing pressure pushed that kink along the tube’s length. That caused the pressure to decrease, which enabled a new kink to appear and the cycle to repeat. “We were super excited because we saw this self-sustaining, periodic, asymmetric motion,” Comoretto told Ars.

The first reason for Comoretto’s excitement was that the flapping tube in his lab was driven by the kind of airflow physics that Peter Marshall, Doron Gazit, and Aireh Dranger harnessed to build their famous dancing “Fly Guys” for the Olympic Games in Atlanta in 1996. The second reason was that asymmetry and periodicity he saw in the tube’s movement pattern were also present in the way all living things moved, from single-celled organisms to humans.

Comoretto’s team decided to build a robot that harnessed the Fly Guys’ physics to achieve natural, almost lifelike movement. But it was harder than it seemed.

Tubular legs

“The movements of these dancing Fly Guys are designed to be random to make them dance in a compelling way,” Comoretto explains. “In our robot, we control the motion through constraining the geometry in very specific ways.” The design comprised a 3D-printed body with four attached tubes bent at the bottom to form the robot’s legs. The frequency of oscillations in the tubes, and therefore the robot’s speed, was regulated by adjusting the amount of air pumped into each tube.

In a standard multi-limbed robot, the remaining challenge would be to synchronize the limbs to achieve different gaits. Surprisingly, though, limb synchronization in Comoretto’s robot appeared as an emergent property of its design.

“Imagine the experiment with metronomes that start to synchronize when you put them on a movable plate—it’s exactly that,” says Johannes Overvelde, AMOLF researcher and co-author of the study. The metronomes synchronize because the moving plate they’re on works as a coupling, connecting them together. In the robot, the same kind of coupling was achieved by connecting all tubes to the same input airflow, which enabled them to communicate through variations in pressure. And it worked wonders.

The robot could adjust its gait to the environment it was in. On the ground, all limbs autonomously activated in synchrony, and the robot ran a bit like a stotting gazelle. In the water, where the tubes met less resistance than on the ground, the limbs moved in antiphase to enable swimming like a dog.

There are still issues to iron out before robots relying on this kind of seamless body-environment dynamics make it out of the early research phase, though.

Into the wild

The first problem with Comoretto’s robot was power consumption. The initial design that his team tested in their lab was tethered, and the air was pumped to the limbs through a thin hose. Fifteen standard liters of air per minute enabled the robot to move quickly and perform all its amazing tricks, but the pump that supplied it used 85 watts. That was too much to even start thinking about dropping the tether and using onboard power sources.

To solve this, the team started by reducing the number of moving limbs from four to two and changing their design to decrease the pressure necessary to form kinks and move them along the tubes. This led to cutting the required power to just 0.06 watts per limb, which enabled a new, untethered robot to move using its own air pump, powered by a small lithium-ion battery. This version used two simple light sensors as its “eyes,” connected to a system that could selectively activate each of the oscillating limbs. It could autonomously go from a dark room to a brighter one or follow an operator carrying a light source.

The most important remaining challenge, though, is to figure out how to control the robot’s behavior. “Now when it hits a wall, it starts to turn left. If it lands in water, it starts to swim backwards. We didn’t come up with that—it just happens,” Overvelde says. “We understand the system but need a better grasp of how to design specific functionalities.” And this grasp will be needed for any applications that require something more than tiny robots that can run, swim, or both.

The goal for the team is to build robotic systems that adjust to the environment using just physics, which would require less computing power (or no computing power at all) to work. “In another project, we are working on a soft, artificial heart—a nice example of a system that interacts with our bloodstream, our blood pressure, and adapts automatically. And when you have an artificial heart, you just want it to work. You don’t want to get software updates,” Overvelde says.

Science, 2025: 10.1126/science.adr3661

Source

Hope you enjoyed this news post.

Thank you for appreciating my time and effort posting news every day for many years.

News posts... 2023: 5,800+ | 2024: 5,700+ | 2025 (till end of April): 1,811

RIP Matrix | Farewell my friend