The Granulobot

February 14, 2024
Maureen Searcy

Physicists Baudouin Saintyves and Heinrich Jaeger develop a modular robot with liquid and solid properties.

Photo of blue-tinted school of fish

Schools of fish, colonies of bees, and murmurations of starlings exhibit swarming behavior in nature, flowing like a liquid in synchronized, shape-shifting coordination. Through the lens of fluid mechanics, swarming is of particular interest to physicists like Heinrich Jaeger, the Sewell Avery Distinguished Service Professor in Physics and the James Franck Institute, and JFI research staff scientist Baudouin Saintyves, who apply physics principles to the development of modular, adaptive robotics.

While the biology of swarming is still mysterious, a swarm’s ability to flow like liquid, act in concert without a leader, and react to its environment inspired Saintyves and Jaeger’s Granulobot. Developed in collaboration with Matthew Spenko, professor in the Department of Mechanical and Aerospace Engineering at the Illinois Institute of Technology at Chicago, the prototype is described in a paper published in Science Robotics.

Granulobot units

The “granular robot” is a collection of simple, cylindrical, gear-like units, outfitted with two magnets that can rotate around the cylinder’s axis. One magnet rotates freely while a battery-powered motor drives the other. This design allows the individual units to connect magnetically and once coupled, push their neighbors and cause them to spin. The contact between each unit moves the aggregate as a whole, much like a swarm, and it can split apart, reassemble, and reorganize to adapt to its environment. And depending on its configuration, a Granulobot aggregate, affectionately called a “blob” by Saintyves, can act like either a rigid solid or a flowing liquid.

The aggregate system “blurs the distinction between soft, modular, and swarm robotics,” says the team. The researchers are working with the Polsky Center to commercialize the technology, which represents the latest stage in the Jaeger lab’s work on soft robotics.

Soft machines

“The field of soft robotics is particularly interesting for applications where robots interface with humans,” says Jaeger. “You don’t want people to get hurt.” Yet the necessity for soft robotics extends beyond safety into suitability. A robot that can change shape can crawl into “nooks and crannies,” says Jaeger, or manage uncertain terrain—both useful for search and rescue, for instance.

For a robot to change shape and perform different functions, its ability to fluctuate between rigid and soft predictably and reversibly is key. Granular materials possess inherent properties that make this transformation possible. This class of materials can transition between liquid and solid behavior based on contact rather than temperature.

That transition is caused by a phenomenon called jamming, which happens when particles in a disordered, chaotic system are so close together that they push against each other, and their flow stops. Jaeger—a condensed matter physicist—describes driving on a highway. Sometimes you’re cruising along, but sometimes you hit bumper-to-bumper cars, and traffic grinds to a halt. When this happens in a granular material, says Jaeger, “it’s essentially a big traffic jam.”

universal gripper made from coffee grounds

Jamming can be seen in action with a brick of vacuum-sealed coffee. Break the seal and the coffee grounds can pour out. Ground coffee works so well in this regard that Jaeger used it to create a soft robotic gripper that can grasp and hold objects regardless of their shape.

A Granulobot cylinder is far bigger than a coffee ground, but the principle is the same. “Jamming is the foundation for the Granulobot to be able to transition from a malleable, more liquid behavior,” says Jaeger, “to something much more like a solid.”


When Saintyves joined UChicago in 2019 following postdoctoral work at Harvard and MIT, he brought experience on how patterns and self-organization emerge in nature as well as robotics, a combination he was already exploring as a visiting artist in JFI and artist in residence at MANA Contemporary.

Self-organization is a process in which a system that continuously receives energy can spontaneously arrange itself into stable states without external direction, arising from interactions between individual components within the system. When energy is produced by “active” components, those local interactions can lead to the emergence of useful global features and behaviors. This phenomenon “is quite ubiquitous in nature,” says Saintyves. Consider living cells: they can autonomously organize into tissues, and through communication among these cells, specific tissue functions emerge with a remarkable robustness. This coordination can happen without explicit instructions from the brain. And of course, this characteristic can also be seen in the movement of swarms.

Saintyves’s autonomous robotics and self-organization interests drove the Granulobot’s design toward enabling the units to self-assemble in aggregates that can transform as continuous materials, self-coordinate locomotion, and react to information and perturbation, all in a decentralized manner—with no “brain” directing their behavior.


The Granulobot is designed to demonstrate the team’s modular, self-organizing approach, but in the future, perhaps the modules could be extremely small—thousands of units so tiny that the group appears to be a singular mass, notes Jaeger. “Another direction that could be really fun to think about is to make them much, much bigger.”

Physics often relies on specific conditions, says Jaeger—extremely small or hot or cold. “Many of my colleagues must work in certain environments, otherwise their whole physics won’t work. The same can be said for life.” Yet the physics principles underpinning the Granulobot are not tied to scale or temperature. “They could work underwater; they could work in outer space,” says Jaeger.

The Granulobot promises exciting advances in robotics,* but Saintyves and Jaeger are physicists. They are using this research to also find new ways to think about matter. “Depending on the self-coordination and the transfer of energy around the environment, your system will either be a programmable material or an autonomous robot. That’s a continuum,” says Saintyves. But “we’re blurring the frontier between matter and robotics.” Within a classical programmable matter approach, the material is a machine. “Here we are exploring the idea that the machine is a material.”

Citation: “A self-organizing robotic aggregate using solid and liquid-like collective states.” Saintyves, Spenko, Jaeger, Science Robotics, Jan 24, 2024.

Funding: National Science Foundation, EFMA-1830939; Army Research Office, W911NF-22-2-0109.

* Saintyves cites the T-1000 shapeshifting, liquid-metal android from Terminator 2: Judgment Day as a sci-fi analog for the Granulobot’s ability to exhibit both solid and liquid properties. He also cites Marvel’s Sandman, whose body is essentially compressed sand. Both villains, he acknowledges. “We’re taking ideas from bad guys and making good things,” says Jaeger. “Hopefully.”

Related News

Faculty, Research, Newsclips