Researchers create world’s smallest autonomous robots

Researchers at the University of Pennsylvania and the University of Michigan have created what they’re calling the world’s smallest fully programmable, autonomous robots—and these things are genuinely impressive.

Each robot measures roughly 200 by 300 by 50 micrometres, which makes them smaller than a grain of salt. That’s barely visible to the naked eye. And yet, despite being absolutely minuscule, these little machines can sense their surroundings, make decisions on their own, and swim around for months at a time. Oh, and they cost about a penny each to make.

“We’ve made autonomous robots 10,000 times smaller,” says Marc Miskin, an assistant professor at Penn Engineering. “That opens up an entirely new scale for programmable robots.”

Can swim and think

The breakthrough here is that these robots don’t need any external control. Previous microscale robots typically required magnetic fields or some kind of tether to operate, but these can function completely independently. They carry their own tiny computers, sensors, and solar panels—all crammed onto a chip that’s a fraction of a millimetre.

Getting robots to work at this scale is genuinely tricky. The physics down there are completely different from what we experience. Gravity and inertia matter less, while drag and viscosity dominate. Miskin describes it like this: if you’re small enough, pushing through water feels like pushing through tar.

So the team had to get creative with movement. Instead of using tiny limbs that would break easily, these robots generate electrical fields that push ions in the surrounding liquid. Those ions then move water molecules around the robot’s body. It’s like the robot creates its own current to ride on.

The brains come courtesy of David Blaauw’s lab at Michigan, which holds the record for the world’s smallest computer. The challenge was running everything on just 75 nanowatts of power, which is actually over 100,000 times less than what a smartwatch uses. The team had to develop special circuits that operate at extremely low voltages to make this work.

These robots can even communicate by doing a little dance. When they need to report temperature readings, they encode the data in their movements, which researchers then decode through a microscope. It’s basically the same principle honeybees use to share information with each other.

The potential applications are pretty exciting. These could eventually monitor the health of individual cells or help construct microscale devices. And since each robot can be programmed separately with its own unique address, you could theoretically have swarms of them working together on different tasks.

“This is really just the first chapter,” Miskin says. The foundation is there now. A platform that can be upgraded with new sensors, faster movement, and more complex programs. That’s a pretty solid start for something you can barely see.