What Holds Up The Robot Takeover?

Mechanical limitations:

The two biggest challenges to making general-purposes robots are, as they always have been, hardware and software. Neither challenge is insuperable, but both are harder than one might think. On the hardware side, there are now lots of robots that can do incredibly cool things. One robot runs faster than the fastest human, another dances Gangnam style. Still another, PR2, folds towels and fetches beer. The catch is that, at the moment, each new robot is like a proof of concept. The ones that are fast and physically powerful, like AlphaDog, a quadruped robot, and the headless but amazing PETMAN, are, for now, still dependent on hydraulic actuators powered by industrial-strength pumps and gasoline engines; they work fine in a laboratory-test environment, but you wouldn’t want one roaming around your home. Others, like Baxter and PR2, are capable of fairly sophisticated movements, but at speeds that are still too slow to be practical around the home. It might take five minutes just for PR2 to grab you a beer. Computer processors keep getting faster and faster—roughly doubling every eighteen months, the rate predicted by the so-called Moore’s Law—and memory gets cheaper and cheaper. But the motors and actuators that move robots aren’t improving nearly as fast. (Battery technology, too, is key, moving quickly, but not quite keeping pace with Moore). In the words of Erico Guizzo, the robotics editor at the IEEE Spectrum, “Lots of people have been working on humanoid robots for decades, but the electric motors needed to drive a robot’s legs and arms are too big, heavy, and slow. Today’s most advanced humanoid robots are still big hulking pieces of metal that are unsafe to operate around people.”

That lets me breathe a little easier.  However, I'm disappointed it will take 5 minutes for the robot to fetch my beer.