Imagine a swarm of microrobots—tiny devices a few hair widths across—swimming through your blood vessels and repairing damage, or zipping around in computer chips as a security lock, or quickly knitting together heart tissue. Researchers at the University of California, Berkeley, Dartmouth College, and Duke University have shown how to use a single electrical signal to command a group of microrobots to self-assemble into larger structures. The researchers hope to use this method to build biological tissues. But for microrobots to do anything like that, researchers must first figure out a good way to control them.
“When things are very small, they tend to stick together,” says Jason Gorman, a robotics researcher in the Intelligent Systems Division at NIST who co-organizes an annual microrobotics competition that draws groups from around the world. “A lot of the locomotion methods that have been developed are focused on overcoming or leveraging this adhesion.”
So far, most control methods have involved pushing and pulling the tiny machines with magnetic fields. This approach has enabled them to zoom around on the face of a dime, pushing tiny objects or swim through blood vessels. However, these systems generally require complex setups of coils to generate the electromagnetic field or specialized components, and getting the robots to carry out a task can be difficult.
Bruce Donald, a professor of computer science and biochemistry at Duke, took a different approach, developing a microrobot that responds to electrostatic potential and is powered with voltage through an electric-array surface. Now he and others have demonstrated that they are able to control a group of these microrobots to create large shapes. They do this by tweaking the design of each robot a little so that each one responds to portions of the voltage with a different action, resulting in complex behaviors by the swarm.
“A good analogy is that we have multiple, remote-controlled cars but only one transmitter,” says Igor Paprotny, a post doctorate scientist at UC Berkeley and one of the lead researchers on this work, which he presented last week at a talk at Harvard University. During his talk, he passed around a container holding a wafer die the size of a thumbnail. On it were more than 100 microrobots.
“What we do is slightly change how the wheels turn,” he says. “Simple devices with a fairly simple behavior can be engineered to behave slightly different when you apply a global control signal. That allows a very complex set of behaviors.” The robots contain an actuator called a scratch drive, which bends in response to voltage supplied through the electric array. When it releases tension, it goes forward, in a movement similar to an inchworm’s. But the key to the robots’ varying behavior is the arms extending from the actuators. A steering arm on a microrobot snaps down in response to a certain amount of voltage, dragging on the surface and causing the robot to turn. By snapping the arm up and down one or two times a second, the team can control how much a given robot turns. To control a swarm, the team designed each robot with an arm that reacts differently during portions of the voltage signal. Computer algorithms vary the voltage sequence, prompting the robots to move in complex ways.
“Electrostatic robots have an advantage in that their power is supplied through an electrode array that the microrobots sit on,” says Gorman. “It can be very compact. Therefore, electrostatic microrobots can be embedded inside other things [like computer chips]. For magnetic robots, you have to supply electromagnetic field, and that requires a larger set-up.” Others have worked on electrostatic microrobots, he adds, but this work is the furthest along.
“His research is very advanced in terms of controlling multiple microrobots,” says Zoltan Nagy, a roboticist at ETH Zürich who works with groups of magnetically controlled robots called Magmites.
“Most of the work to date has been on controlling a single robot that can move around in a pre-defined area on a substrate,” adds Gorman. “However, many of the applications of interest will require control of lots of robots, like a colony of ants.”
So far, Paprotny has been able to control up to four robots on a single surface at once, and the robots can move several thousand times their body length per second, as detailed in a paper that is currently submitted for review. His next plan is to adapt the setup for a liquid environment so that the microrobots can assemble components of biological tissue into patterns that mimic nature.
“We’re trying to come up with ways of self-assembling tissue units,” says Ali Khademhosseini, an associate professor at Brigham and Women’s Hospital at Harvard Medical School and a specialist in tissue engineering who is collaborating with Paprotny. “In the body, tissues are made in a hierarchical way—units repeat themselves over and over to generate larger tissue structures.” Muscle tissue, for example, is made from small fibers, while liver tissue has a repeating hexagonal shape.
Khademhosseini has encased cells in jelly-like hydrogels and assembled them (using methods that include liquid-air interactions and surface tensions) into different regions to mimic biological tissues. But he thinks the self-assembling microrobots will allow more control in creating the tissues.
“We can try to combine cells and materials in microfabrication systems to come up with structures and assemble them in particular ways using the techniques Igor has developed,” says Khademhosseini.
He envisions fabricating the gels and cells on top of teams of robots working in parallel to construct different parts of a tissue. “We could use the robots to do assembly,” he says. “The cells, once they’re assembled, come off from the robots, letting cells rearrange further to make things that are indistinguishable from natural tissue.” Initially, he hopes to create small patches of heart tissues, and then things like heart muscles and valves, and assemble them all together in a heart. “That’s where things are heading,” he says. “But right now the challenge is we’re still not very good at making each of these individual components.”
Become an MIT Technology Review Insider for in-depth analysis and unparalleled perspective.Subscribe today