Forget Legs — This Robot Has Wheels That Walk, Skate, and Ride Like a Motorcycle
RAI Institute's 15kg RoadRunner switches between walking, inline skating, and motorcycle mode. Built by the founder of Boston Dynamics.
While everyone's debating whether humanoid robots can sort packages or play tennis, Marc Raibert's team just asked a different question: what if a robot's wheels could do everything legs can — and faster?
What Happened
RAI Institute (formerly The AI Institute), founded by Boston Dynamics creator Marc Raibert, unveiled RoadRunner on March 23 — a 15-kilogram bipedal wheeled robot that can seamlessly switch between three locomotion modes. In side-by-side mode, its wheels sit parallel like a cart for stability. Flip to inline mode, and it rides like a motorcycle for speed. Need to climb stairs? The wheels become feet, stepping up one at a time.
The demo video racked up over 272,000 views in two days, showing RoadRunner standing up from the ground, skating across flat terrain, climbing and descending stairs, and balancing on a single wheel. The robot's legs are entirely symmetric — knees can point forward or backward — giving it an uncanny range of motion.
The most impressive technical detail: a single control policy handles both side-by-side and inline driving, trained entirely in simulation and deployed to real hardware with zero additional tuning. That zero-shot sim-to-real transfer is something most robotics teams still struggle with.
Why It Matters
RoadRunner isn't competing with humanoids — it's sidestepping them. At 15 kg, it's a fraction of the weight of robots like Figure 03 (135 lbs) or Atlas. For use cases where speed matters more than dexterity — warehouse patrol, industrial inspection, search and rescue — a wheeled biped that can also climb stairs could be more practical than a full humanoid.
RAI Institute hasn't announced pricing or availability. This is still a research platform, not a product. But coming from the same mind that built Spot and Atlas, RoadRunner hints at a future where the most useful robots aren't the ones that look most human — they're the ones that move most efficiently.
