The guy who built this robot is the same person who created xenobots in 2020, the first living robots made entirely from frog cells. That trajectory tells you everything about where robotics is heading.
Sam Kriegman spent years studying how biological organisms survive damage. A starfish loses an arm, grows it back. A flatworm gets cut in half, both halves become new worms. Biology solved the resilience problem a billion years ago: make every piece a complete agent.
He took that principle and made it mechanical. Each module in this metamachine has its own motor, battery, and computer. Cut the robot in half and you get two robots. The severed leg rolls away on its own and can rejoin later. The remaining body recalibrates its gait instantly and keeps walking.
The AI designing these bodies is the part that should make every robotics company nervous. Kriegman's algorithm runs simulated Darwinian evolution, breeding thousands of body configurations, keeping winners, discarding losers. The designs it produces look nothing like any robot a human engineer would sketch. Three-legged things with tails. Five-limbed creatures where limbs double as spines. Forms that move like seals, lizards, and kangaroos depending on terrain.
These metamachines ran outdoors across gravel, mud, sand, tree roots, and uneven brick. They jumped obstacles, did aerial spins, and flipped themselves upright when knocked over. No retraining. No recalibration. Zero sim-to-real gap.
Boston Dynamics spends years hand-engineering a single quadruped body plan that breaks when you remove a leg. Kriegman's AI generates thousands of body plans in hours, and the ones it picks are functionally immortal.
The 2020 xenobots were biological cells on a petri dish. Five years later, the same researcher has athletic machines built from Lego-like blocks running through mud outdoors. The compression from living cells to modular hardware to AI-evolved locomotion happened in one lab, in one researcher's career. That's the pace now.
Northwestern University researchers developed modular robots using AI that can adapt to damage and navigate unpredictable terrain, according to a new study