ED-209

Gordon McComb wrote:


ED209. I

Hi Gordon. I'm not sure I can say anything specific about inherent advantage of such a leg design, but bird legs in general, as well as ED-209, have similar geometry to the back legs of all "toe-walkers", such as dogs, cats, deer, horses, and most other quadrupeds. Birds look kind of funny, but I think they just evolved that way, ie with the same basic rear-leg design as other vertebrates.
In essence, they all walk with their "heels" up off the ground and have an elongated foot-segment which is analogous to the segments including the soles of human feet. All vertebrate-quadrupeds have strong rear-ends, and the heel-up geometry both increases their strides and also adds a large amount of springiness to the step, which is a great advantage during running, for shock absorption and energy recovery. You will notice that was one of the advantages cited recently regarding the passive-dynamic walkers - that they are 10X or more energy efficient than the Asimo type bots. Animals are said to have upwards to 70% energy recovery in their strides. Energy is stored in the muscles and tendons during foot-fall, and recovered during the subsequent push-off. I kind of see the ED-209 geometry as a bipedal way of taking advantage of all of this.
You can see most of what I am talking about by taking a look at the running-dog page .... notice how the legs are coming down in frame #15 of the galloping dog ...
http://www.oricomtech.com/projects/leg-run.htm
Regards flamingos balancing on 1 leg, I think that's a different thing. Basically, during balancing, their legs are straight. However, if you look at the body design of birds in general, in addition to the bipedal dinos like T-rex, what you immediately notice is the body weight is distributed evenly on both sides of the vertical leg posts. They couldn't really work any other way. Eg, the long neck+head on one end of the T-rex is balanced by the long tail on the other end. Birds are similar, except that almost all of their weight is in the body mass, but it's distributed evenly fore-n-aft. Take a look at some ostriches for instance. Huge round bodies, evenly-centered, and pencil-thin necks, etc.
http://images.google.com/images?q=ostrich
Back to ED-209, bascially, I think the design is naturally based upon the geometry of bird + quadruped rear-legs [ie, backward-pointing hocks], and the low forward leaning body is simply a way to balance the weight over the legs. I suspect this design is a LOT easier to balance than an upright biped, like humans. I've been playing with simple papers designs for a biped like this too. However, I think you should be able to do it with much simpler leg designs than the usual 12-servo biped or the lynxmotion Scout.
[no short answers today, I guess ;-)].
- dan michaels www.oricomtech.com ====================
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
My Lynx Scout looks promising for this.
I have ~30 90 Watt motors set aside for a project. I was going to do a big Shelob, A big hex, with 3 axis hips, or Scorpion, but I thought about doing an ED209 as well.
One of the things I have against the biped designs out there, is only 6 axis in the leg, where I belive that a 3 axis ankle would give a distinct advantage. Crouding all those motors together in the hip and ankle is a bit of a bother. I had a '209 like design with an articulated femur and tibia. Each would rotate on it's axis. The math is a bit of a bear, and the geometry is not completely practical, but it does offer a different way of doing things.

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
blueeyedpop wrote:

a big

doing
6 axis

a bit

tibia.
way of

I happened to read the article on Hexapod Hoedown in Sept. 2004 Nuts&Volts mag last night. Shelob looked interesting, but seems that Phil Davis' isobots.com site is down now. What I found most interesting were the comments about adding more and more DOFs to the legs in order to get rid of foot "scuffing" on the floor. I wonder if it's really worth going to all that trouble, except possibly as an engineering challenge. Nature didn't make us perfect either - when my shoes wear out, I go buy new ones. Actually, I thought that possibly adding a little rotator cuff would have greatly reduced any scuffing on the hexapod feet.
What I agreed with most in the article was the idea of using sensors on the feet along with feedback to control the gait, especially as this is the way nature solved the locomotion problem. Nature apparently doesn't calculate inverse kinematics. It makes a predictive estimate, and then compensates in real-time. IOW, may be a lot easier to compensate for a non-perfect design than to expend the effort to try and make a perfect design. Different philosophies, I guess. I perceive that engineering and comp.sci students are taught slightly different viewpoints in school.
have fun, - dan michaels =================
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
blueeyedpop wrote:

a big

doing
Well, looks like my post from yesterday isn't gonna show up, so try again ....
Couple of nites ago, I happened to read the Hexapod Hoedown article in Sept 2004 Nuts+Volts mag. First, it looks like Phil Davis' isobots.com site regarding the Shelob robot is gone already. Also, I found it interesting how much emphasis was placed in the article about foot shuffling. I can't really imagine this is a big problem with most hexapods since they can lift their legs straight up. If it's a problem with the grounded legs rotating, I figure one could put a rotator cuff on the feet. Also, humans have this problem too, and I just buy new shoes after a while.

6 axis

a bit

tibia.
way of

I guess now you're talking about a 5 or 6 DOF hexapod leg, which adds even more to the complexity. It's a good engineering challenge though, both mechanically and regards building a controller.
The one thing I agreed with a lot from the article was the idea of using sensing + feedback for refining the leg movements in real-time. Animals don't calculate inverse kinematics, rather they make predictive estimates for movements, and then use proprioceptive, touch, and other feedback to modify the movements in real-time. So, I've often wondered whether it's really such a good idea to go to all the trouble of doing inverse-kinematics at all, as opposed to something a lot simpler using closed-loop feedback - especially for a small robot.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
After washing around a bunch of ideasw for the next project, i have (tentitively ) decided on either a single legged bouncer, or the holonomic drive on a sphere trick. ( see CRM about a year ago )
The ball balencer is something that i have been ruing over for 1/2 a decade now, and with a waterjet cutter at work, I think I can take a reasonable stab at it now.

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
blueeyedpop wrote:

I've also been working on a bouncer...the application for those 12V electric impact wrenches at Harbor Freight I mentioned. Only mine has three legs, not one, partially because I want to make the software easier. It's more of a frog hopper than a bouncer.
Now the only trick is figuring out how to keep it from rattling itself loose.
-- Gordon
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
I think it is going to have to be a ball roller thing this go around

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
blueeyedpop wrote:

They're fun. I bought a basic version of the ball roller at Dick Smith Electronics in Redwood City (when DSE was in America) in 1986. The family pets just *adored* it!!
-- Gordon
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Gordon McComb wrote:

around
Smith
I must have missed the thread about the holonomic ball roller, but I know of one ball-bot ... not sure if it's what you're talking about ... see Par_cho|r
http://visionenkessel.de/projekte.htm
Also, Nasa has a few hopper/jumpers around .....
http://www.space.com/businesstechnology/technology/robofrogs_000914.html http://www.roboticsonline.com/public/articles/archivedetails.cfm?id#1
I'm thinking along the opposite line to Mike - trying to figure out how to make legs less complex instead of more complex. With reference to the passive-dynamic bipeds that hit the news a few weeks ago, with the right design you can decrease energy use in a leg by 10X or so. The P-D walkers have too few DOFs, and I doubt any of them could ever get up after falling over, but there must be some lessons to learn from energy-efficient leg designs.
- dan michaels ==============
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
imagine a holonomic drive platform on top of thr ball
a 2 axis pendulum...

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.