CMU's Ben Stephens is lucky enough to have a Sarcos humanoid to play with, and play with it he has, using a motion capture system to teach the robot to dance, if you want to call what the robot's doing dancing. There's some serious researchy stuff going on also, though: while dancing, the robot manages to not fall over, dynamically keeping its balance while coming as close as possible to replicating the captured human dance movements.
In unstructured environments (like any environment where humans are allowed to run around), balance is a big issue for robots, since they never know when they may accidentally get shoved by a wayward human. And it's important that the robot be able to deal with being shoved, partially for the sake of the complicated and expensive robot, but also for any small children and/or pets who may find themselves underneath a robot with inadequate balancing skills. To this end, Ben has been teaching the Sarcos robot to deal with a push in the same way that humans do: by taking a step forward to keep its balance:
If you've ever watched humanoid hobby robots do just about anything, you've probably noticed that they're fairly horrible at keeping their balance. Let's hope that research like this eventually trickles down to the consumer level, if for no other reason than to make humanoid soccer and humanoid kung-fu competitions a little more interesting.
Evan Ackerman is a senior editor at IEEE Spectrum. Since 2007, he has written over 6,000 articles on robotics and technology. He has a degree in Martian geology and is excellent at playing bagpipes.