The robot is 1-DOF. It starts by pumping energy into the closed system until he is rotating fast enough to be able to lift off above the height of the bar, spin 360 and return with hands on the bar.
see second 57 in slow motion:
the fingers open and he leaves the bar while center-of-mass (CM) has almost only vertical upright velocity.
the upright velocity of the CM is enough for the body to take off and reach just a couple of centimeters above the height of the bar.
the opening of the fingers is timed so perfectly that the fingers leave the bar just 1/1000 of a second after CM has passed the bar's height level and has pure vertical upright speed direction. by leaving an instant after reaching vertical upright speed, the body gets not only upright motion but also spins.
The spin is at a rate such that when it completes 360 degrees, the fingers are back on the bar. ie: the spin is timed with the return back to bar-level from the short flight up.
I am very curious to hear from him if he has actually done a full system identification and modeled the robot in order to feed the parameters into matlab and achieve the perfect pumping and timing... or was this just rial and error?
Seems from videos on his page that the robot's actuation system was designed from scratch (http://www.youtube.com/user/hinamitetu)Statistics: Posted by limor — Sun Jan 17, 2010 10:23 pm
]]>