Artificial Spacetimes for Microrobot Control

Physicists have long understood how to focus and guide light, even with limited control over each beam’s initial trajectory. Here we show these same concepts can be repurposed into powerful techniques to steer and control robots. Specifically, we demonstrate a correspondence between ray optics, general relativity, and gradient control that allows us to convert optical components like lenses and waveguides into fields that robots can use to navigate, converge, and traverse complex paths. Our approach only requires local actions from the robot (e.g. measuring the control field) yet produces predictable behaviors with global properties such as obstacle avoidance and guaranteed transit times. Finally, we validate our optics-inspired approach in the lab by guiding microrobots through complex environments and executing prescribed motions with no knowledge of the robot’s initial heading or location.