The video basically showcases how indefatigably unstoppable the thing is, and the difference between "unstoppable machine" and "unstoppable killing machine" is a gun attached to the end. That would sound less paranoid if the video didn't clearly and prominently state this is a DARPA project. Imagine it busting in your door at 3AM and lobbing in a couple frag grenades while telling you in a monotone synthesized voice that you're "liberated. "

That's actually the main thing I wanted to chime in: It's DARPA, which means even if a private university or civilian research team is doing the research it's still a DoD project about as civilian as the Skunk Works. Most engineers just want to create things, efficient, productive things. Many don't care if what those things do so efficiently is kill. Most of the Pentagon knows both of these things and takes full advantage of each. It's been half a century since the eggheads produced an invincible super weapon, and since Rummys plan to launch orbital kill satellites in the late '70s didn't go anywhere, they need to earn their keep.

It is VERY cool that servers and gyros keep it on its feet no matter what. I might've fallen myself going from snow to pavement with patchy ice (I still say its knee touched down at 1:47, so under college rules it's down right there. ) IIRC that was the main reason the Pentagon said Battlemechs weren't feasibly when they looked at them seriously a generation ago: The actuators and gyros needed to keep a 70 ton robot upright and moving would weigh 200 tons. FASA always had them running on fusion, but then, FASA always had them shooting lasers; something like the machine gun packing exoskeleton in the Matrix (or Hitlers at the end of Wolfenstein 3D) could be very feasible, and then the question is "how close to the front do you need your robots pilots to be?" One day "fog of war" may be reduced to lag from spy satellites and control headsets.