One Small Leap for a Muscle, One Giant Step for Robotics

These robots have real muscles, and they’re finally making big moves. Researchers at ETH Zurich have developed muscle actuators that flex strongly enough to enable small robots to jump, swim, walk and grasp.  

In brief 

• Instead of developing them in separate processes, ETH Zurich researchers have, for the first time, jointly designed centimetre-scale living muscle tissue and its soft artificial scaffold on a computer. 

• The resulting muscle actuators achieve a range of motion up to eleven times greater than that of earlier systems with a comparable amount of muscle tissue. 

• Using these muscle actuators, the researchers built small biohybrid robots capable of a range of movements, including jumping, swimming, walking and grasping. 

Lab-grown skeletal muscle tissue can contract, but the movement is often so slight that it is barely visible. This poses a problem for biohybrid robots designed to move using living muscles. Researchers at ETH Zurich have now demonstrated that it is not just the amount of muscle tissue that matters, but above all how the muscle and the artificial scaffold interact.

Their bioactuators combine living skeletal muscle tissue, grown from mouse muscle cells, with a soft hydrogel scaffold. This soft, water-rich material bends and stretches like living tissue. What is new is that both parts are designed together as one system. The researchers first used a computer simulation to determine where and how much muscle tissue to place and what properties the hydrogel scaffold should have.

Aiste Balciunaite and Mike Y. Michelis, co-lead authors of a external page study recently published in Nature Communications, jointly developed a new approach for this purpose. The muscle and the scaffold are designed together from the outset. Balciunaite, a doctoral student in the Soft Robotics Lab led by ETH Professor Robert Katzschmann, explains: “Instead of specifying a fixed form for our muscle geometry, we wanted to design the whole thing from scratch on the computer.”

Muscle movement visible to the naked eye

To do this, the researchers designed the muscle tissue configuration and the scaffold simultaneously on the computer, rather than one after the other. An algorithm searched for configurations in which the contraction of the muscle produced the greatest possible movement of the entire component. The researchers then built the computer-designed muscle actuators in the laboratory and tested them. The scaffolds were 3D-printed with light using a technique called xolography, which takes about one minute per scaffold. 

“Of course, it was an iterative process,” Balciunaite recalls. The experiments provided information on material properties and manufacturing conditions, which was fed back into the computer simulation.

The result was the creation of what the researchers call bioactuators, each about 1.5 centimetres long. These actuators consist of living muscle tissue and an artificial hydrogel scaffold. They can flex so strongly that the movement is visible to the naked eye. The researchers report a range of motion up to eleven times greater than that of earlier systems with a comparable amount of muscle tissue. “With previous systems, you sometimes had to look under a microscope to see the flexing,” says Balciunaite.

Jumping, swimming, grasping

The newly developed bioactuators can jump, swim, move across a surface and grasp objects. What impressed Balciunaite the most was the jumping. Jumping, she says, is an unusual movement that requires a particularly strong change in shape. For her, this experiment shows how important the interaction between the living muscle and the soft artificial scaffold is.

With a single contraction, the actuator lifts itself about five millimetres off the bottom of the dish, a third of its body length. A walking version covers 6.2 millimetres per second, faster relative to its body length than any walking biohybrid robot reported so far.

These small robots can move so powerfully not just because of their muscle tissue. It is crucial that the muscle and the scaffold form a seamless, firm connection. Fine grooves in the scaffold help the tissue anchor firmly and align its muscle fibres. This ensures that as little muscle strength as possible is lost during the transfer to the scaffold.

Leaving the lab is still a long way off 

For now, however, biohybrid robots depend on a nutrient solution that keeps their muscle tissue alive. Researchers are still a long way from developing a robot that can be used over the long term. Katzschmann explains that larger systems require an artificial blood supply: a vascular system that delivers oxygen and nutrients to the tissues and removes waste products. This is one of the biggest challenges for future development.

Another challenge is the natural variability of living tissue. Unlike industrially manufactured components, cultured muscle actuators do not all behave exactly the same way. The researchers therefore aim to better understand and control how muscle tissue develops in the laboratory as it matures.

Why use living muscles at all? 

For Katzschmann, the study shows above all the potential of the interaction between living tissue and soft artificial structures. Muscles can do more than generate force. They can change their shape, thereby influencing how a body moves and flexes. The muscle shape itself becomes something engineers can design, rather than a fixed constraint to work around.

At the same time, the study shows how closely different fields of research need to work together to achieve this. Only through the interplay of cell biology, materials science, robotics and computer-aided design was it possible to develop muscle tissue and an artificial scaffold as a single integrated system. 

For Katzschmann, this also offers longer-term potential: in the future, robotic systems could make greater use of the characteristics of living organisms, rather than relying solely on conventional motors and rigid components for movement. “We’re just beginning to build robots in a more natural way,” he says.

Reference

Balciunaite A, Michelis MY, Filippi M, Paniagua P, Yasa O, Katzschmann RK: Co-optimized volumetric muscle design for large dynamic deformations in versatile biohybrid robots. Nature Communications 17, 10732, 11. September 2026, DOI: external page 10.1038/s41467-026-77655-1