View Full Version : Ankle question

10-14-2011, 03:31 PM
so from what I have seen in the forums so far is the ankles are hte most important servo since it typically fails first. I was wondering if someone can give me a quick rundown on how to find the needed torque for ankles (and other important joints). I dont mean standard FBD static calcs as those are straight foward but in standard just standing ankles dont seem to be a problem if the feet balance the load on each side of the servo. I am more interested in the dynamic calcs as it has been a while since I have worked with mechanical systems most of my work is with air flow calcs etc. Thanks for any help guys I do appreciate all the info I have gotten from you.

EDIT: pictures would be helpful if possible.

10-14-2011, 03:42 PM
Just wondering... Are you now building a biped instead of a quad?

10-14-2011, 04:19 PM
No. I am just curious and also I am considering a 4DOF quad now. I am just wondering because I am not sure how much the dynamic solutions will affect a quad as the static solutions are very simple using FBD's. I would think if you move two legs of a wuad simultaneously you would have to do similar calcs.

P.S. Don't think I have to time to jump straight to a biped and learn using that. If I had the expenses I would probably start with a hex as the gaits will be simplier but having a budget kinda ruins that as it will add at least $250 in just AX-12's.

10-14-2011, 04:32 PM
The "foot" servos (as I think of them) on a 4DoF quad seem like the ones doing the least work to me. And quads really aren't hard, gait-wise... (I'm also biased and find hexes to be boring)

10-14-2011, 05:58 PM
The dynamic solution depends on what your motion profile is. so... the load depends totally on your motion profiles and geometry, weights, ect.

The quads with 4 degrees of freedom that have been built here (immortal and Numa) pretty much just have the foot always staying vertical. These servo's see hardly any loads at all relative to the other servo's.

The reason bipeds have issues with ankle servo's is due to how the simple walking gaits are constructed.


Most hobby robots move through a more or less series of static positions to generate their walking movement. This means they shift the weight of the robot to be over the foot and then lift the other leg up, or at least pretty close to that. The problem people have is that the "x" distances are too long, or/and that the shifting of body weight creates a huge lever arm that the ankle has deal with if the acceleration is too much. The taller robots face this problem even more as "h" is increased.

The reason robots like Darwin can hold higher payloads (besides the better servos) is you can tune the movement to be more dynamic. The program itself doesn't calculate these dynamic forces, but since the movement is parametric you can tweek it much easier and better then a pose based robot. When it walks, it isn't holding static positions. Darwin also has the legs very close to each other so that the x distance is quite short. This means you need smaller feet of course, which is also hard to deal with on pose based robots.