Picture this: Three people walking in different directions reach a sidewalk corner at the same moment a skateboarder whizzes up. Who goes where, and when?
The social rules for getting around each other are both complicated and largely unwritten, a serious problem as more autonomous vehicles (cars, robots, etc.) hit the streets too. Computer science professor Silvio Savarese may have a solution.
He and his lab have been recording traffic on Stanford’s campus via drones and feeding that footage to computers to study how people move en masse. Based on that research, they’ve written an algorithm that allows a robot to “read” a sidewalk, calculate who is moving where, and—based on patterns it has observed in the past—decide how best to move without awkwardness or incident. They’re now testing their code with Jackrabbot, named for the nimble animals scampering around the Farm.
So fear not: Robots may still take over the world someday, but at least they won’t bump into you on the sidewalk.
