Ok i have worked with the SES version for this leg design.
As i mentioned reverse swing control, I will be calling this configuration of leg design RSC-3v1. "Reverse Swing Control - 3(DOF) version 1.
http://i531.photobucket.com/albums/dd355/innerbreed/100_0476.jpg
I will construct the the other three legs sometime this week, only if i can bring myself round to taking parts from other project! 
SILENT THOUGHT…
new video!!!
It has 72×92pixel CMOS camera,and recognize many ball’s color(red,yellow,and green)and potition(in X-Y plane).
The arm has 4-joint parallel link. Therefore, it is designed for no dependence of the hand-direction, the camera always face down.
His last video was awesome.
And the leg design are really smart and “torque saving”.
Good luck with your SES version of this quad!
If you can make it walk like in that video without using wheels, then I’ll be impressed 
But then again, I’m not really sure if it’s possible for movements to be that smooth on a quad without cheating a little bit 
Maybe one of these days I’ll try with my own… If I ever get any free time 
Can’t agree less, it would be really impressive if the robot able to walk without the wheels.
Anyway, if you watch his video closely, you will notice that the quadruped cannot switch his walking gait without actually settling down onto the ground and reposition all the legs. What I mean by switch gait is when the bot change walking direction; be it forward-reverse, forward-rotate, forward-turning etc.
yes this gait changing idea doesnt really work for me.
i have seen and also theaded a few vids of omni-directional quad’s so i know its possible.
not sure if it can work with this design of leg but i will find out! Its still a quality design even with this Drop-reposition gait!
I will be buildng the second leg later using spare parts i have, and while doing this i will be thinking about if i want to take my current project apart to complete the build!
thanks for the support and advise!
Very nicely done! This might make a great advanced kit. 
8-Dale
I will be converting the kit into SES. Working on the legs and trying to think up the best dsign for a body! i could use the Quadrapod Body Kit - Mini
img370.imageshack.us/img370/3794/qbk01im2.jpg
As said earlier there is a video of this quad with Omni-directional gait.
although you can see the shift in the COG and the Tripod stance created by the legs in order to keep it stable, we have to remember it is a simulation so we shouldnt always trust what we see.
Check out in the said video, 0.09 to 0.19 (transition of curvature gait to straight gait) and 0.28 to 0.38 (transition of straight gait to rotational gait). The transition gait is actually much stable than the straight/curvature/rotational gait.
In principle, the best transition COG is generally the middle point of present and sequential COG coordinate.
In hexapod (considering the tripod gait), it has 2 set of independent tripod per cycle. (Ripple and wave gait can actually be extrapolated from tripod gait). Therefore, hexapod (tripod) gait transition is almost instantaneous, as the transition COG can be computed easily. Theoretically, hexapod can change gait on the move, meaning without having to return to the home (or starting) leg position/configuration. Xan’s Phoenix tripod gait has clearly demonstate such behaviour, though I’m not sure if he use the same principle.
Whereas in quadruped, it has 4 set of dependent tripod. The subsequent triangle (tripod) has 2 of its corner points (and 1 edge of the triangle) are of the previous tripod. During transition, the best transition COG is computed as the earlier statement: the middle point of present and sequential COG coordinate. And if we use the home (or starting) position principle, it takes 2 cycle/transition, or more acurately 8 steps/transition. We don’t see this in hexapod tripod gait because of the independency of its tripod.
Ideally, a quadruped is desired not to have any home (starting) position. If so, the quadruped need to know upfront the leg configuration of the next walking gait. Computational wise, this is already very complicated.
For fluid motion, a special stability and transition algorithm is necessary, which by far, still unsolved.
innerbreed…
tnay…
I have added your comments to my discusion on “Gait algorithm study”
lynxmotion.net/viewtopic.php?p=45093#45093
You know that you can put the person’s name in the brackets and you’ll get a nice little box that says:
instead of
{quote=“innerbreed”}{/quote}
like that, but with non-squiggly brackets
