Researchers at the University of Pennsylvania and the University of Michigan have built the smallest fully programmable autonomous robots ever created. These microscopic machines can swim through liquid, sense their surroundings, respond on their own, operate for months at a time, and cost about one penny each to produce.

Each robot is barely visible without magnification, measuring roughly 200 by 300 by 50 micrometers. That makes them smaller than a grain of salt. Because they function at the same scale as many living microorganisms, the robots could one day help doctors monitor individual cells or assist engineers in assembling tiny devices used in advanced manufacturing.

Powered entirely by light, the robots contain microscopic computers that allow them to follow programmed paths, detect local temperature changes, and adjust their movement in response.

The work was reported in Science Robotics and Proceedings of the National Academy of Sciences (PNAS). Unlike previous tiny machines, these robots do not rely on wires, magnetic fields, or external controls. This makes them the first truly autonomous and programmable robots at such a small scale.

"We’ve made autonomous robots 10,000 times smaller," says Marc Miskin, Assistant Professor in Electrical and Systems Engineering at Penn Engineering and the papers’ senior author. "That opens up an entirely new scale for programmable robots."

Why shrinking robots has been so difficult

Electronics have steadily become smaller over the past several decades, but robotics has not followed the same trajectory. According to Miskin, independence at sizes below one millimeter has remained an unsolved challenge. "Building robots that operate independently at sizes below one millimeter is incredibly difficult," he says. "The field has essentially been stuck on this problem for 40 years."

At everyday scales, motion is shaped by forces such as gravity and inertia, which depend on an object’s volume. At microscopic sizes, however, surface-related forces dominate instead. Drag and viscosity become overwhelming, dramatically changing how movement works. "If you’re small enough, pushing on water is like pushing through tar," says Miskin.

Because of this shift in physics, conventional robotic designs fail. Small arms or legs tend to break easily and are extremely difficult to manufacture. "Very tiny legs and arms are easy to break," Miskin explains. "They’re also very hard to build."

To overcome these limitations, the researchers developed a completely new way for robots to move that works with the physics of the microscopic world rather than fighting against it.

How microscopic robots swim

Fish and other large swimmers move by pushing water backward, generating forward motion through Newton’s Third Law. The tiny robots take a very different approach.

Instead of bending or flexing, the robots generate an electrical field that gently pushes charged particles in the surrounding liquid. As those ions move, they drag nearby water molecules with them, effectively creating motion in the fluid around the robot. "It’s as if the robot is in a moving river," says Miskin, "but the robot is also causing the river to move."

By adjusting this electrical field, the robots can change direction, follow complex paths, and even coordinate their movement in groups that resemble schools of fish. They can reach speeds of up to one body length per second.

Because this swimming method uses electrodes with no moving parts, the robots are remarkably durable. According to Miskin, they can be transferred between samples repeatedly with a micropipette without damage. Powered by light from an LED, the robots are able to keep swimming for months.

Packing intelligence into a microscopic body

True autonomy requires more than movement. A robot must also be able to sense its environment, make decisions, and power itself. All of those components must fit onto a chip that is only a fraction of a millimeter across. This challenge was taken on by David Blaauw’s team at the University of Michigan.

Blaauw’s lab already holds the record for creating the world’s smallest computer. When Blaauw and Miskin met at a Defense Advanced Research Projects Agency (DARPA) presentation five years ago, they quickly realized their technologies complemented each other. "We saw that Penn Engineering’s propulsion system and our tiny electronic computers were just made for each other," says Blaauw. Even so, turning that idea into a working robot required five years of development.

One of the biggest obstacles was power. "The key challenge for the electronics," Blaauw says, "is that the solar panels are tiny and produce only 75 nanowatts of power. That is over 100,000 times less power than what a smart watch consumes." To make the system work, the team designed specialized circuits that operate at extremely low voltages, cutting power consumption by more than 1000 times.

Space was another major constraint. The solar panels take up most of the robot’s surface, leaving very little room for computing hardware. To solve this, the researchers redesigned how the robot’s software works. "We had to totally rethink the computer program instructions," Blaauw explains, "condensing what conventionally would require many instructions for propulsion control into a single, special instruction to shrink the program’s length to fit in the robot’s tiny memory space."

Robots that sense and communicate

Together, these advances produced what the researchers believe is the first sub-millimeter robot capable of real decision-making. To their knowledge, no one has previously placed a complete computer with a processor, memory, and sensors into a robot this small. That achievement allows the robots to sense their environment and respond independently.

The robots include electronic temperature sensors that can detect changes as small as one third of a degree Celsius. This capability allows them to move toward warmer regions or report temperature values that can serve as indicators of cellular activity, offering a way to monitor individual cells.

Communicating those measurements required an inventive solution. "To report out their temperature measurements, we designed a special computer instruction that encodes a value, such as the measured temperature, in the wiggles of a little dance the robot performs," says Blaauw. "We then look at this dance through a microscope with a camera and decode from the wiggles what the robots are saying to us. It’s very similar to how honey bees communicate with each other."

The same light that powers the robots is also used to program them. Each robot has a unique address, allowing researchers to upload different instructions to different units. "This opens up a host of possibilities," Blaauw adds, "with each robot potentially performing a different role in a larger, joint task."

A platform for future microscopic machines

The current robots are only the starting point. Future versions could carry more advanced programs, move faster, include additional sensors, or function in harsher environments. The researchers designed the system as a flexible platform, combining a robust propulsion method with electronics that can be manufactured cheaply and adapted over time.

"This is really just the first chapter," says Miskin. "We’ve shown that you can put a brain, a sensor and a motor into something almost too small to see, and have it survive and work for months. Once you have that foundation, you can layer on all kinds of intelligence and functionality. It opens the door to a whole new future for robotics at the microscale."

The research was conducted at the University of Pennsylvania (Penn) School of Engineering and Applied Science, Penn School of Arts & Sciences, and the University of Michigan Department of Electrical Engineering and Computer Science. Funding came from the National Science Foundation (NSF 2221576), the University of Pennsylvania Office of the President, the Air Force Office of Scientific Research (AFOSR FA9550-21-1-0313), the Army Research Office (ARO YIP W911NF-17-S-0002), the Packard Foundation, the Sloan Foundation, and the NSF National Nanotechnology Coordinated Infrastructure Program (NNCI-2025608), which supports the Singh Center for Nanotechnology, along with Fujitsu Semiconductors.

Additional co-authors include Maya M. Lassiter, Kyle Skelil, Lucas C. Hanson, Scott Shrager, William H. Reinhardt, Tarunyaa Sivakumar, and Mark Yim of the University of Pennsylvania, and Dennis Sylvester, Li Xu, and Jungho Lee of the University of Michigan.