ROBOTS AS MATERIALS?
With the help of embryonic-inspired fluidization processes, a proof-of-concept robot collective has the ability to mimic different material properties. New research from the University of Florida suggests that vision may play an important role in helping people learn new motor skills.
Written by Nancy Kristof
Multiple robots in this system will fluidize to create new shapes and forms. Photo: Matthew Devlin/University of California, Santa Barbara
Inspired by previous research that demonstrated how living embryonic tissues can fluidize themselves into different shapes, a research team from the University of California, Santa Barbara and Technische Universität Dresden in Germany set out to test if and how those processes might be translated into robotics. The idea was to create a multi-robot system that could maintain stiff and strong forms while still being able to fluidize to take on new shapes, move in a new direction, or even heal itself.
“If you can figure out how the embryo forms itself in a robotic system, that unlocks all sorts of imaginations,” said Matthew Devlin, lead author of the paper and a former doctoral researcher in the lab of UCSB mechanical engineering professor Elliot Hawkes.
Devlin received his undergraduate degree in biomedical engineering before shifting to robotics when pursuing a doctorate in mechanical engineering. After taking a class on self-organization led by Otger Campàs, a former UCSB professor and now the director of the Physics of Life at TU Dresden, Devlin wrote his final paper on vibrating multi-robot systems that could work together like embryos do, an idea that caught the attention of Hawkes, his advisor.
How the robot collective operates and moves. Video: Matthew Devlin/University of California, Santa Barbara
Both Campàs and the research team’s “simulation genius,” Sangwoo Kim, formerly of UCSB and now at EPFL in Switzerland, are trained physicists who quantitatively understand developmental biology, bringing invaluable interdisciplinary knowledge to the team, Devlin said.
“There was just enough overlap that we could effectively work together. And I think that's really where the magic is,” he explained. Through simulations, the team was “able to figure out the type of force pattern that the embryos were likely applying onto each other that resulted in them elongating,” Devlin continued. They approximated it as a circle within a quadrant, with forces varying from clockwise to counterclockwise motions to change the direction of the elongation.
It took the team less than a year to figure out the simulations, but nearly three times as long to develop the physical robots robust enough to demonstrate it at scale, finally landing on an individual robot size of 70 millimeters.
“It kind of felt like we were testing a biological system that we didn’t fully understand,” Devlin said.
The team tested hundreds of physical robot iterations against simulations, which provided insights into how to best recreate embryonic force fluctuations. Embryonic cells individually jiggle and both pull and push on each other to reach a fluidized state.
“In embryos, they have actomyosin and all these really cool cellular structures to push and pull, but we replicate that just with gears that spin on motors.”
–Matthew Devlin, former doctoral mechanical engineering researcher at the University of California, Santa Barbara
“It turns out, if you increase the amplitude of those fluctuations, it results in the entire collective being more likely to flow or rearrange,” Devlin said.
That intercellular push and pull is one of the three high-level mechanisms required to get robots to act like embryos that can change shape and strength.
“In embryos, they have actomyosin and all these really cool cellular structures to push and pull, but we replicate that just with gears that spin on motors,” Devlin explained. The robots include eight motorized gears along the exterior, moving in a direction commanded by their environments.
Secondly, the robots need to be able to adhere to each other. In embryos, cellular adhesion allows for the transmission of biochemical positions that allow them to connect to each other and share movements. This was replicated in the robots via magnets incorporated into small chambers around the robot’s perimeter.
The robots also needed polarity to replicate the signals that an embryo uses to distinguish its head from its tail, which the team accomplished by installing light sensors on top of the robots. By shining a flashlight with a certain polarity onto 20 robots, then “they can all line up in a single file line without talking to each other, without communicating at all, without knowing anything about the world around them, besides which way is up,” Devlin explained.
The robots come together to form a bridge, highlighting their various abilities to self-heal structures and act as tools. Video: Matthew Devlin/University of California, Santa Barbara
“That’s all it takes to make the system, and it can be very powerful,” he said, adding that simplicity is key.
The study of force fluctuations on the robots’ movement also provided new insights that may be applied back to biology. Devlin said that almost on a whim, they tried applying fast and slow speeds to the gears, which provided insight into how force fluctuations lowers the mean power required, though it may take longer.
“We hypothesize that biological cells are likely a power-limited system,” said Devlin, who noted that the robotic models may make it easier for the team to test its own hypotheses and use the system for things such as quantitative measurements for calorimetry.
A robot collective may also be useful in limited resource applications such as space, Devlin noted. The system could benefit hardware platforms, where the robot collective is used to flow underneath and lift or support a heavy weight. It’s more proof of why interdisciplinary research is so important, added Devlin, pointing out that there is a wealth of research from biology and chemistry that still needs to be translated.
“We all speak math the same,” Devlin said of the overlaps in engineering disciplines. “That’s really where the magic is.”
Nancy Kristof is a technology writer in Denver.
© 2025 The American Society of Mechanical Engineers. All rights reserved.