LittleDog Learns Several Terrifying New TricksS

The last we checked in with LittleDog, the smaller, scrappier version of BigDog and frequent resident of my worst nightmares, we took some solace in the fact that he wasn't quite as nimble as his predecessor. Now: terrifyingly nimble!

Yes, it's true. LittleDog, which is being developed at USC as part of Darpa's robot locomotion initiative, will now chase you across rocky terrain, over crevices, and up stairs of various heights. Any weaknesses we might've perceived in the initial videos of LittleDog have vanished and with them our hopes of ever resisting the RoboDog takeover.

This newest version of LittleDog is even programmed to teach itself to be more efficient (read: deadly), possessing the ability to evaluate the difference between a good foothold and a bad one and adjusting its steps accordingly. While this thing's getting smarter all the time, my dumb human programming remains the same: when you see a RoboDog, turn around and put one foot in front of the other, as quickly as possible. [YouTube]