Engineers at the University of California San Diego have created a four-legged soft robot that doesn’t need any electronics to walk — just a constant source of pressurized air.

The team, led by Michael T. Tolley, a professor of mechanical engineering at the Jacobs School of Engineering at UC San Diego , detailed the air-walker technology in the Feb. 17, 2021 issue of the journal Science Robotics .

A lightweight system of pneumatic circuits, made up of tubes and soft valves, handles the air from an onboard CO2 cannister. The softer approach eliminates the need for bulky circuit boards and pumps traditionally found on flexible, elastomeric robots.

The valves act as oscillators, controlling the order in which pressurized air enters the air-powered muscles in the robot’s four limbs. The three valves create a high-pressure state throughout the air-powered circuit, with a delay at each inverter. By selectively restricting the injection of air, the robot has an almost turtle-like gait. (See for yourself in the Tech Briefs TV video below.)

The robot turns, too, thanks to simple mechanical sensors — little soft bubbles filled with fluid. The bubble sensors are placed at the end of booms protruding from the robot’s body.

When the bubbles are depressed, the fluid flips a soft valve in the robot that switches the direction of the limb rotation and causes a reverse in direction.

The robot can walk on command or in response to signals it senses from the environment.

Each of the robot’s four legs has three degrees of freedom powered by three muscles. The legs are angled downward at 45 degrees and composed of three parallel, connected pneumatic cylindrical chambers with bellows.

When a chamber is pressurized, the limb bends in the opposite direction. As a result, the three chambers of each limb provide multi-axis bending required for walking.

Dylan Drotman working on a robot
Dylan Drotman, working on a soft robot back in 2017. (Image Credit: : UC San Diego Jacobs School of Engineering / David Baillot)

In the future, the UCSD researchers want to improve the robot’s gait so it can walk on natural terrains and uneven surfaces.

"This work represents a fundamental yet significant step towards fully-autonomous, electronics-free walking robots," said Dylan Drotman, a Ph.D. student in Tolley’s research group and the paper’s first author.

In a short Q&A with Tech Briefs below, Drotman explains more about where this type of robot can be used, and how it can potentially do much more than just walk.

Tech Briefs: How does the robot stop and start?

Dylan Drotman: The robot started walking forward when the circuit was pressurized, and the robot stopped walking when the circuit was depressurized. We achieved this by opening or closing a valve connected to the source of pressurized gas.

Tech Briefs: How is the air pressure maintained?

Dylan Drotman: The pneumatic control circuits were powered by pressurized gas stored in a pressure vessel, either a small onboard CO2 canister, or a larger offboard compressed air tank connected by a tube. In both cases, we used a pressure regulator to maintain the desired circuit pressure. The pneumatic circuits were designed to oscillate in response to this constant pressure input, generating the patterns of output gas pressures needed to power the muscles of the legs for walking.

Tech Briefs: The video (above) shows how the robot uses a boom to kind of "bump" and sense its surroundings. Are there other ways that the robot could respond to the environment?

Dylan Drotman: The touch sensor was used as a simple demonstration but anything that can press on the control valve in the robot's "brain" that switches the walking direction could be used as an input. Our touch sensor transmitted a force applied at the end of the boom to the brain through a transmission fluid. While this type of physical contact is the most direct way to switch the state of this valve, one could imagine a device that generates pressure through a chemical reaction, or a smart material that generates pressure in response to specific stimuli like light, heat, or the presence of a chemical compound.

Tech Briefs: How are commands given?

Dylan Drotman: The sequential behaviors were mechanically “programmed” based on the design of the pneumatic circuits and a small number of inputs either from touch sensors or a manual controller. The pre-programmed circuit generated the rhythmic walking motion, while the inputs chose the walking direction. One of the supplemental videos in our publication  (SI Movie S2) shows us directing the robot with the manual controller, while another (SI Movie S5) shows the robot changing direction based on the input from a touch sensor. Aside from these inputs, everything else is controlled onboard by the pneumatic brain.

Tech Briefs: Could it be programmed to perform other actions than walking?

Dylan Drotman: The soft ring oscillators themselves could be used for any application that requires periodic motion, like swimming, running, rowing, or mixing. One could then use a similar approach as we did to adjust the rhythmic motion (like change the direction or speed) in response to simple inputs as discussed above.

We see potential applications for cases where simple, electronics-free robots are needed that can generate rhythmic motion with a small number of control inputs. Examples include some medical devices, human-safe robots (like toys), and robots for exploration or search-and-rescue.

Tech Briefs: How could the robot be used in other applications, like an MRI or mine shaft?

Dylan Drotman: The only metallic components on our current prototype are the pressurized air tank and pressure regulator. One could imagine a device that has these components outside of an MRI machine connecting to the robot with only a tube, or it may be possible to design non-metallic versions of these components for an untethered device.

Due to the absence of electronic components, in an environment with risk of spark-ignition hazards (a mine shaft, a grain silo, inside a fuel tank), we believe our prototype walking robot may present a lower risk than a robot powered by electronics. However, we have not yet tested our system for any of these applications, and design optimizations would almost certainly be required depending on the application. In addition, we believe that the pneumatic circuits in this work could be used to control low-cost robotic toys. Since the circuits can be made completely out of molded or 3D printed polymers (rather than materials like metals and semi-conductors), we anticipate reduced materials and manufacturing costs.

What do you think of the air-powered robot? Share your questions and comments below.