AI & Computing3 min read
MIT Engineers Build Paper-Thin Robot That Swims Using Light-Activated Living Muscle Cells
A two-fin robot powered by a single layer of engineered muscle could lead to cheaper, self-healing machines for delicate environments that rigid hardware might damage.

MIT engineers have built an extremely thin swimming robot powered by living muscle cells that contract when hit with light. The device, described in a new study in Advanced Functional Materials, swam and steered through a simple underwater maze, which the researchers say is the first demonstration of a very thin, two-dimensional robot that moves from place to place using living muscle tissue.
The robot has a flexible gel framework roughly the length and width of a stick of gum. The gel forms two fins, each coated with a layer of muscle cells far thinner than a human hair. The cells have been genetically modified to contract when exposed to light. Shining light on a fin makes it flap and pushes the robot through the water. By changing which fin is lit and how often, the team controls the robot's speed and direction.
"You can think of the robot as having two independent muscles," said Ritu Raman, associate professor of mechanical engineering at MIT and a study author. "If we shine a light on just one, only that muscle moves. If shining on both, they both flap."
In tests, the researchers placed the robot in a large petri dish of water and moved a light source by hand to guide it through a maze. At top speed it covered about four body lengths per minute. That is far slower than an Olympic swimmer, who can travel up to 65 body lengths per minute, but comparable to a cow shark cruising leisurely, according to the source. "It takes a lot of force to move through water versus air," Raman said. "The robot's quite strong, given its size."
Most existing biohybrid robots, including earlier ones from Raman's group, rely on bulky, three-dimensional chunks of lab-grown skeletal muscle that require millions of cells. Raman said thinner designs like this one could be cheaper to build and could move more efficiently.
The work builds on an artificial muscle disk the lab created last year, inspired by the human iris. That disk used a gel patterned with grooves, on which muscle cells grew into a thin layer that contracted in multiple directions under light. But the cells moved only about 100 microns, which Raman described as tiny from a robotics perspective.
To get more force and movement, the team reworked the gel foundation. The earlier disk used fibrin, a very soft gel that could shrink quickly under the pull of contracting cells. The researchers tuned three features: the gel's chemical makeup, its stiffness, and the size and shape of the grooves that organize muscle growth.
They replaced fibrin with gelatin methacrylate, or GelMA, a material commonly used to engineer tissue. Stiffer GelMA formulations worked best, with cells growing in better alignment and contracting more strongly. Square-bottomed grooves aligned cells better than longer, curved-bottom channels. Aligned cells are more likely to merge into fibers that contract in a coordinated way. A gel film about half a millimeter thick struck a balance, sturdy enough to support a single layer of muscle yet light enough that the contracting cells did not pull away from the surface. The team also "exercised" the muscle with repeated light pulses before building the robot, which strengthened the tissue.
The researchers say the robot's current shape is simple, and the main goal was to show that an extremely thin layer of engineered muscle can produce enough force to swim. Next, they plan to optimize the body design for faster swimming. In the source, Raman suggested that even slow muscle-powered swimmers could be used for environmental monitoring in aquatic settings. She also said biohybrid robots could one day explore environments too fragile or unpredictable for conventional hardware, because living tissue is soft, responsive to its surroundings, and can heal itself.
The source describes these applications as possibilities rather than demonstrated capabilities. The robot was tested only in a petri dish, with light directed by hand. The study's other authors are first author Maheera Bawa, Arielle Berman, Laura Schwendeman, Ferdows Afghah, and Seanbiron Johnson. The Office of Naval Research provided partial funding.