Inspired by nature: A baby fawn can walk within seven hours of its birth. Until it can use the legs successfully, it throws around all four limbs in a highly adorable, although frenetic, way to figure it all out. Robots, on the other hand, need people to be on hand to pick them up if they fall over.
An important advancement: Google researchers got a four-legged robot to learn to walk forward and backward, and turn left and right, completely on its own. It took a few hours, but it did it without anyone intervening.
How it works: Reinforcement learning is commonly done in simulation – a virtual doppelgänger of the robot flails around a virtual doppelgänger of the environment until the algorithm is robust enough to operate safely. It is then imported into the physical robot.
The drawback: This method requires a very precise, easy-to-model physical environment where the robot can move around. It relies on a motion capture system placed above the robot to determine its location. That won’t be possible in the real world!