It’s long been possible to make extremely small robots, but they usually need some form of direct external control just to operate. Cornell scientists may have solved that problem on a basic level, however. They’ve created microrobots (no more than 250 micrometers across) with basic electronic “brains” that let them walk autonomously. Two- and six-legged robots move relatively simply, while a four-legged “dogbot” changes speed when an operator sends laser pulses.
The trick was to build a complementary metal-oxide-semiconductor (or CMOS, as computer enthusiasts know it) clock circuit whose signal produces phase-shifted square wave frequencies that set the gait of the robot’s platinum-based legs. Photovoltaics control both the legs and the circuit. The design is far from complex at just 1,000 transistors (for context, a GeForce RTX 4090 has 76.3 billion), and it’s still large enough that it effectively serves as the robot’s body. However, even that is an achievement — the exceptionally low power demands saved Cornell from having to use relatively gigantic photovoltaics.
These inventions are a far cry from the more sophisticated large-scale autonomous robots you see today. They can move forward, but not much else. The researchers see this as just a beginning, though. They believe future microrobots could be crucial to healthcare, where they could perform internal surgery and clean your arteries. Elsewhere, they could detect chemicals and eliminate pollutants. Any such bots are likely years away, but this project suggests they’re technically possible.