MIT builds an ultra-thin swimming robot powered by living muscle and steered with light

MIT engineers have unveiled a small swimming robot powered by a single layer of living muscle cells, much thinner than a human hair. The work was published in the journal Advanced Functional Materials and was publicized through ScienceDaily on October 23, 2026. This piece summarizes the material received, which is a full press release and not a paywall teaser.

The device consists of a flexible gel frame, roughly the length and width of a stick of gum, that forms two fins. Each fin is coated with a layer of muscle cells genetically modified to contract when exposed to light. Shining light on one fin or the other makes it flap, and the robot moves forward through the water. By varying which fin is lit and how often, the team controls speed and direction.

In experiments, the robot swam and turned through a simple maze inside a large water-filled Petri dish. The researchers moved the light source by hand to steer it. At top speed, it covered about four body lengths per minute. According to the release, that is far less than an Olympic swimmer (up to 65 lengths per minute) but comparable to the leisurely pace of a cow shark. Ritu Raman, associate professor of mechanical engineering at MIT and an author of the study, notes that moving something through water takes much more force than through air and that the robot is quite strong for its size.

According to the authors, this is the first demonstration of a very thin two-dimensional robot that moves using living muscle tissue. Most current biohybrid robots rely on bulky three-dimensional structures of lab-grown skeletal muscle that require millions of cells. Raman argues that thinner designs could be cheaper to make and move more efficiently.

The project builds on an earlier achievement by the lab: an artificial muscle disc inspired by the human iris, presented last year. In that work, radial and concentric grooves were etched into a gel, and the cells grew along those channels, contracting in different directions when exposed to light. It showed that very thin muscle layers with complex architectures could be grown, but the movement was only about 100 microns, minuscule from a robotics standpoint.

To achieve greater force and displacement, the team redesigned the tissue scaffold. The earlier disc used fibrin, a very soft gel that tended to shrink under the force of the cells. The researchers explored three variables: the chemical composition of the gel, its stiffness, and the dimensions and geometry of the grooves that guide muscle growth.

They compared several groove shapes, including narrow square-profile channels and longer channels with curved bottoms. The cells aligned better in the square grooves, which matters because aligned cells fuse more easily into fibers and form stronger, more coordinated muscle. They also replaced fibrin with gelatin methacrylate (GelMA), a common material in tissue engineering, and adjusted its formulation to achieve different stiffnesses. The stiffer gels gave the best results, with better alignment and more powerful contractions.

As for thickness, a film of roughly half a millimeter struck a good balance: firm enough to support a single layer of cells and light enough to keep the contractions from peeling off the tissue. In addition, the muscle underwent a kind of training with repeated pulses of light before being integrated into the robot, which improved its performance.

In the final robot, both sides of the gel have square-bottomed grooves and are covered with cells that fuse into organized fibers. Raman sums it up as two independent muscles: if only one is lit, only that one moves; if both are lit, both flap.

The robot's current shape is deliberately simple, since the goal was to prove that an extremely thin layer of engineered muscle can generate enough force to swim. The next step is to optimize the body to swim faster. Raman suggests that, even at low speed, a muscle-powered swimmer could be used for environmental monitoring in aquatic settings, or to explore environments too fragile or unpredictable for conventional hardware, taking advantage of the fact that living tissue is soft, responsive to its surroundings and able to heal.

The research was funded in part by the US Office of Naval Research. The team includes Maheera Bawa as first author, along with Arielle Berman, Laura Schwendeman, Ferdows Afghah and Seanbiron Johnson. It is worth remembering that this is a laboratory proof of concept: low speed, manual steering with light and applications that remain hypothetical.