Imagine designing a machine whose moving parts have to be kept alive. Before asking what it can do, you have to ask what it needs to survive.
That question gives biohybrid robotics its unusual character. The engineering challenge extends beyond shape, wiring, and control. A successful design has to make room for the needs of biological tissue while turning its movement into something useful.
What the researchers built
ETH Zurich researchers combined lab-grown mouse skeletal muscle tissue with a soft hydrogel scaffold. They co-designed the tissue arrangement and scaffold using computer simulations, then fabricated and tested the resulting bioactuators. Different configurations demonstrated jumping, swimming, walking, and gripping. [1, 2]
The muscle tissue is alive. That does not establish that the complete machine is an organism, has a brain, or experiences anything. Movement alone is no evidence of consciousness.
It is tempting to give a twitching machine an animal personality. A more revealing way to watch it is as an engineered partnership: one component supplies contraction, while another determines how that contraction changes the shape of the whole device.
Watch the muscle do the work
These are three separate experimental demonstrations. The jump happens in liquid, from the bottom of a dish; it should not be mistaken for a robot leaping across a dry laboratory bench.
Original supplementary movies 1, 4, and 8 from the study. Credit: Balciunaite et al., Nature Communications (2026), CC BY 4.0. Videos compressed for web playback without changing their sequence; preview frames extracted by WolfieWeb. Some clips have no audio.
Why the scaffold matters
The advance is the relationship between the muscle and its support. An optimization algorithm searched for arrangements that turn contraction into larger deformation. The team reports up to eleven times the range of motion of earlier systems with comparable muscle volume. That is a movement comparison, not eleven times the strength. [1, 2]
Think of the difference between pulling on a loose piece of string and pulling on the same string attached to a carefully arranged mechanism. The force matters, but so does everything that happens between the pull and the result. Here, that relationship is an essential part of the machine.
Small movements, useful context
| Measurement | Reported result | Context |
|---|---|---|
| Actuator length | About 1.5 cm | Centimetre scale |
| Jump rise | About 5 mm | In liquid |
| Walking speed | 6.2 mm/s | A walking configuration |
Source: ETH Zurich’s report and the original study. These measurements describe particular experiments, not a universal specification for every device. [1, 2]
A few millimetres may sound modest. For this kind of research, the useful questions are about control and repeatability: does a predicted deformation happen, does the device interact with its surroundings, and can another specimen perform in a comparable way?
The catch: the machine needs life support
The tissue currently depends on nutrient solution. Larger systems would need ways to supply oxygen and nutrients and remove waste; biological variability also makes consistent performance difficult. ETH explicitly describes long-term practical operation as a distant goal. [2]
This changes the meaning of maintenance. A conventional device can often sit idle until it is needed. A living component brings conditions that must be sustained even while the machine is doing nothing useful. Any future application has to justify the cost and complexity of caring for it.
The experiments demonstrate muscle-powered motion and interaction. They do not establish a self-sufficient household robot, a medical treatment, or a machine that can heal itself after arbitrary damage.
What makes this worth following
Our take: the fascinating question is whether a living component can eventually earn its place in a dependable machine. That would require more than an impressive movement. It would require useful operation, predictable behaviour, and a practical way to keep the tissue functioning.
For now, watch the contraction itself. The future of robotics may include bodies assembled through a mixture of fabrication and cultivation. This study offers a concrete experiment to examine while that possibility is still taking shape.
Study links, dates & media credits
- [1] Peer-reviewed study: Balciunaite, Michelis, Filippi, Paniagua, Yasa and Katzschmann. Co-optimized volumetric muscle design for large dynamic deformations in versatile biohybrid robots. Nature Communications 17, 10732. Published September 11, 2026. DOI: 10.1038/s41467-026-77655-1.
- [2] Institutional report: One small leap for a muscle, one giant step for robotics, ETH Zurich, October 22, 2026.
- Media: Research video frame, Figure 1 and supplementary movies 1, 4 and 8 from the original study. Balciunaite et al., 2026, Creative Commons Attribution 4.0. Images converted to WebP; video preview stills extracted. Research videos compressed for web playback; content and sequence retained.
WolfieWeb interpretation is identified as commentary. Research reviewed November 8, 2026. WolfieWeb is independent of ETH Zurich.