Navigating stepping stones and other uneven terrain with balance is no easy feat. But new algorithms are helping robots do just that.
The quest to make robots appear more human-like keeps on keeping on. Atlas, Boston Dynamics’ high mobility, humanoid robot that is designed specifically to navigate outdoor terrain, just got a little more human-like in how it moves thanks to a new set of control algorithms.
Researchers from the Institute for Human and Machine Cognition Robotics Lab in Pensacola, Fla., created the software that allows Atlas to cross an uneven path of cinder blocks in eerily human-like fashion. The algorithms direct Atlas to move its foot to feel out the ground ahead of it, move forward and course correct its balance if needed by swinging its torso and arms.
In an MIT Technology Review article covering the advancement notes that Atlas explores a new contact surface by shifting the center of pressure around the foot, and “an available foothold is inferred by the way in which the foot rotates about contact edges … during the exploration.” That information also helps Atlas determine how to hold its foot as it takes a step, the article said, and the arm waving helps it maintain and regain balance so it doesn’t fall over—a pretty common occurance with robots attempting to perform human movements.