MIT's Tiny Flying Robot Gets 450% Faster With a New AI Control System

MIT engineers have given their microcassette-sized flying robot a dramatic upgrade, using a new two-part AI control scheme to boost its top speed by roughly 450 percent and its acceleration by about 250 percent compared with the team's previous best demonstrations.

The robot, which weighs less than a paperclip and is powered by fast-flapping artificial muscles rather than motors, can now pull off ten full body-flip somersaults in just eleven seconds while staying within 4 to 5 centimeters of its planned flight path. The new controller balances high maneuvering performance against the tight computational budget such a tiny onboard system can actually support.

Researchers see the agility gains as a step toward practical uses for insect-scale robots — searching earthquake rubble, inspecting tight machinery, or navigating collapsed structures that are too small or dangerous for conventional drones to enter.

Photo of a CIA "Insectothopter" micro-drone via Wikimedia Commons — a historical example of an insect-scale flying robot — released into the public domain by the Central Intelligence Agency.