E pluribus unum - "out of many, one" - is not only a motto for the United States. It's a good credo for microrobots.
A research collaboration between Cornell and the Max Planck Institute for Intelligent Systems has shown how a swarm of microrobots spinning on a water surface can together generate the fluidic torque needed to manipulate passive structures without any physical contact.
This collective behavior was demonstrated to operate gears and move objects, with the aim of eventually performing microscale tasks and biomedical procedures.
The research published Feb. 25 in Scientific Advances. The lead author is Steven Ceron, Ph.D. '22, now an assistant professor at the University of Michigan.
"At small scales, contact-based manipulation can be limiting, and flow-based manipulation offers a great alternative," said Kirstin Petersen, associate professor and an Aref and Manon Lahham Faculty Fellow in the Department of Electrical and Computer Engineering in the Cornell Duffield College of Engineering and the paper's co-senior author. "We're showing that in spite of their size, adding more microrobots creates stronger flows and greater torque transfer."
A major focus of Petersen's Collective Embodied Intelligence Lab is leveraging robots' physical interactions with each other and their environment so the collective can exhibit a large variety of behaviors without advanced control. The microrobots in this case are simple 3D-printed polymer discs, 300 micrometers in size. Because the bots are too tiny to hold onboard computation, the researchers developed a relatively simple way to affect their behavior: the discs are sputter-coated with a thin layer of a ferromagnetic material, placed in a 1.5-centimeter-wide pool of water, and subjected to oscillating orthogonal magnetic fields, which cause each disc to spin on its axis.
"We call them microrobots, but truly the individual agents are just mechanical rafts, and you can think of the whole system as the robot," Petersen said. "When they spin, they create flows, and now the hydrodynamic, capillary and magnetic interactions between individuals change the behavior of the collective.
"That means you're getting these emergent phenomena, where they all interact together and can create a large range of behaviors, far beyond what you would be able to do with robots that were not tightly coupled through their environment," she said.
The lab's collaboration with the Max Planck Institute previously demonstrated the ability of the microrobots to actuate and self-organize into diverse patterns. Now, the researchers conducted a range of experiments to model such behavior. The team deployed different quantities of bots - as few as 10, up to 1,000 - that generated fluidic torque to perform tasks that included: rotating multiple concentric rings; turning circular gears, grippers, rack-and-pinion systems and 3D buoy-like structures; and absorbing and expelling dozens of passive objects.
"Modeling many interacting microrobots at these scales is challenging," Petersen said. "So we combined experiments and simulations to understand how their collective flows and interactions produce these behaviors."
The bot-rafts performed as anticipated, but the researchers were surprised to identify some unexpected activity. When introduced to a comparatively large rotating object, i.e., a couple millimeters long, the bots clumped together and entered a novel crawling state.
"They start out uniformly distributed around the object, but at certain frequencies they cluster on one side and move together as if crawling around its perimeter. It emerges from how the bots' flows interact with each other and with the object's boundary," Petersen said. "It's a behavior that could be useful in the future for controlled transport or positioning at small scales."
With their ability to generate and transmit fluid-driven torque collectively, swarming microrobots may be particularly well suited for biomedical applications that require gentle, distributed manipulation without direct contact.
"We've shown that these microrobots can be added to a wide range of passive structures to actuate them," Petersen said. "Instead of building a larger integrated mechanism, you can use a collective of simple microrobots to drive millimeter-scale elements such as gears or grippers."
Co-authors include Gaurav Gardi, who led the development of the computational model, and Metin Sitti of the Max Planck Institute for Intelligent Systems in Stuttgart, Germany.
The research was supported by the Max Planck Society, the National Science Foundation, the Fulbright Germany Scholarship and the Packard Foundation Fellowship for Science and Engineering.
