I used a 2-axis acceleration sensor. Is located in the head it. It then measures the gravity mainly. The measured value, by the angle of rotation will change me. --- I will zero gravity measurement value measurements of the axis when the axis of the sensor was leveled to the ground ---. Position in the rotation I was divided into four regions. To detect the falling signal rising and in (up, down, left, right) the boundary position. (With a threshold detection value zero gravity fall detection circuit and rising that). I was triggered by a combination of (H, L signal using the threshold as zero gravity detection value) of H and the raw signal, L signal. Based on these signals, and to determine when to stretch or bend the leg.
Well that was cool. But the page doesn't exactly have much in the way of words. Despite the spammy nature of the newslamp.com domain --and despite the spammy nature of the youtube video being in response to Katy Perry's Part of Me music video-- any idea what microcontroller is being used?
EDIT: And now this video is rocketing to the top of reddit, number two at the moment. Video is on liveleak.com. Way to give the actual producer credit, internet.
To be fair, there have been far more interesting posts on HN that were ignored because it was only a YouTube video. Would you be commenting if it wasn't linked to a different site (however spammy looking the domain)?
Of course! I wanted to know the micro-controller used :)
Looks like gilgoomesh has a direct link to the creator's actual youtube account as well, not the embedded rip-off from youtube user chrill22, which looks like a typical view farmer. It looks like chrill22 took hinamitetu's "Quadruple Backflip(NO.16)" video, trimmed out the part where hinamitetu was setting the robot up, and then re-upped it. Absolutely infuriating.
And judging by bradleyjoyce, it looks like there's more technical info to be found, albeit in a language I don't understand.
There are lots more acrobatic feats on his YouTube page:
https://www.youtube.com/user/hinamitetu?feature=watch