Sidewalks are complex environments. Sure, they’re second nature for humans, but the delicate balance of humans on foot, humans on bikes, humans with animals, humans in clusters, and humans on skates, skateboards, or other configurations are a lot for a machine to process. Put that sidewalk on a college campus, where people are walking in all directions, clustering and stopping suddenly, and emerging from buildings, and that’s even more challenging for a robot to navigate. Researchers at Stanford built a machine to learn the rules. And to make it seem extra friendly, they gave it a tie and a straw hat.
Meet the Jackrabbot:
Here it is, rolling around people.
The Jackrabbot is a learning machine. It scans the environment for people and tries to guess where they will move next, extrapolated from testing and learning algorithms. This math is really the big innovation here – studying pedestrian behavior and predicting it is key. The current robot, built as a testbed for this algorithm, is remotely controlled, but future incarnations will be driven by the math itself, navigating sidewalks like humble ancestor of R2-D2.
Watch Jackrabbot amble about below: