A lionfish swims against the current, its tail moving like a pendulum in slow motion. But this fish isn’t like its cold-blooded counterparts. It’s a robot, and instead of blood running through its veins, it circulates an energy-dense liquid to both power its batteries and push its fins. The robot, described today in the journal Nature may be the first step in addressing two major hurdles in robotics—power and control—with one solution. And thanks to the energetic liquid pumping through its pseudo-vascular system, this robot might be a little more like us.

Robots don’t typically work the same way living things do. Instead of an intricate network of multifunctional parts, robots tend to be made of isolated components that each serve a single purpose, explains mechanical engineer Robert Shepherd of Cornell University, principal investigator of the new study. For instance, they might have one system to address power and another to control motion, which isn’t always efficient. By contrast, the human circulatory system is multifunctional: It pumps blood throughout our bodies and by doing so, it also helps regulate our body temperature and transports cells to fight off infections.

There are examples of circulatory systems in nature that are even more efficient than our own. In fact, Shepherd’s initial inspiration for the robo-lionfish wasn’t actually much of a swimmer. Rather, he was fascinated by the high-flying bar-tailed godwit, a migratory bird that he calls a “super athlete.” A godwit can fly for a week without stopping, but first doubles its weight in fat to prep for the flight.

“That really stuck with me that you can add energy to an animal in a multifunctional way—both thermal insulation and storing energy, and then distributing it in a way that’s efficient,” says Shepherd. “If you compare that to our batteries [in robots], they often don’t perform any other function than providing energy and adding weight.”

With this in mind, Shepherd wondered if there was a way to make batteries in robots successfully manage both power and control. Lots of robots already pump hydraulic fluids, such as water, through their systems to apply force that moves some of their parts. If they could replace a typical hydraulic fluid with one that stores energy, he thought the fluid could then do more than just facilitate mechanical motion. Using a multifunctional hydraulic could also save energy in the long-run, since traditional robots with solid batteries often need additional battery packs for long-term operation, which add extra weight and reduce performance.

In their design, Shepherd and his team, who have applied for a patent on their design, used what’s called zinc iodide redox flow batteries, which have a liquid electrolyte solution in them that acts as an energy reserve. The energy-rich liquid contributes to chemical reactions that charge the battery, while also working as a hydraulic fluid that circulates through the lionfish and moves its fins. To allow movement, the fins are made of flexible electrodes and a soft silicone skin. Pumping hydraulic fluid into one side of the tail fin inflates the skin and causes the fin to bend around the stiffer center sections towards the other side. Reversing the direction of the fluid bends the fin the other way, allowing the fish to swim as the fluid oscillates. Pectoral fins are also powered by the fluid, and can fan outwards, mimicking the fin movements lionfish use to communicate.

Placing the lionfish in a salt water tank, the team observed the robot could successfully swim against a current. In experiments, they let the robot swim for up to two hours, but calculated that it could theoretically operate for as long as 36 hours. They also estimated that the robot’s energy performance was about three to four times better than a traditional design using a normal hydraulic fluid like water.

Shepherd explains that the multifunctional use of solid batteries isn’t new. For instance, the batteries in a forklift act as an energy source, while also providing weight to stabilize the machine during heavy lifting. But the diverse use of liquid batteries hasn’t been explored until now. “Now that the idea is out there,” says Shepherd, “We’re hopeful that when people use hydraulics they can ask, ‘Can I replace the hydraulic fluid with electrolytic fluid—does that make sense with the energy cost versus weight for a denser fluid in my system?’”

“The idea of using the liquid as the battery is really great,” says Robert Katzschmann of ETH Zurich, a roboticist who has worked on other robotic fish , but was not involved in this research. However, Katzschmann maintains concerns about the efficiency of the battery, and emphasizes that the concept might be better showcased out of the water, where avoiding the extra weight of solid battery packs becomes critical without the help of buoyancy.

“In theory it’s great, because you could make a robot that’s not underwater,” says Katzschmann. “If you want to make a walking robot, it’s a little more difficult. And no one has shown a fully soft robot that can fly, so it makes sense to show it underwater as an idea, but there’s still a lot of work for them to do.”

Shepherd is optimistic about the battery’s improvement. He emphasizes that the chemistry of their battery is safe to handle but “not as energy dense as it could be.”

“The challenge is increasing the energy density while being safe,” he says. “We know where it can go, but we have to go there more cautiously.” And like Katzschmann, he envisions this work contributing to future robots on land, which could possibly be used in search and rescue missions. “We’ve made a stretchable system, so the form that you’re currently limited to could change,” Shepherd adds. “Certainly, the future is hybrid systems, at least for terrestrial systems…where soft parts are used for sensing and overlaid over electromechanical and fluid actuators.”

While there are many advances to be made in the field of soft robotics, Shepherd’s lionfish suggests that, so far at least, things are moving along swimmingly.