Leg design sketchups:

[Robotboy86]

Attached are two images of the sketchup I have been playing with..  first one has the leg curled under the base, second one has it extended.  Both of these have segment rotation under 60 degrees, and I am hoping once I get a bit more planning done, to start building the structure..  then hook in servo's, batteries, program it and make it move! 

We will see though..


Comments are appreciated

[

[Admin]

Since you asked for comments . . .

I recommend avoiding the use of those gears for two reasons: gear meshing is hard to get right, and gear efficiency will kill ~10% of your servo strength.

I didnt see your battery yet, just remember to keep it low.

I just realized this post was also yours:
http://www.societyofrobots.com/robotforum/index.php?topic=1641.0
(good to keep them together)

I like the way you positioned the legs under the robot - this allows for efficient walking. With your leg design, when the legs are straight vertical your servos require zero torque. Use this to your advantage.

[Hal9000]

I'd try to steer well away from that design if I were you to be honest.

Try modyfying slightly something that has already been done.

[Admin]

I'd try to steer well away from that design if I were you to be honest.

Why?

[HDL_CinC_Dragon]

Well ive been having this idea lately that I havnt been able to fiddled around with much but I guess I have nothing to lose by mentioning it..

My idea was to create a tendon system for the limbs. basically a cable like the ones used for bicycle brakes would be anchored at the tip of the last digit of the limb and then again to an actuator or a servo that winds up the cable. Each limb would require at least 2 but could implement 4 for multiple axis control.

basically creating a finger. at each 90? angle, there would be a cable running through its entirety. Just like a tendon in your finger. That way, instead of having a bulky servo sticking out of the side of the leg, its all concealed in the robots body, hiding behind the curtain pulling all the strings. literally. It works in my mind and im very very confident id be able to do it.... but I simply dont have the resources 

Think of the way muscles work. When you move your arm up, theres 2 muscles working. 1 is pushing, and 1 is pulling. Thats basically exactly what my idea would play on. Maybe one day ill draw up some CAD designs... the blue prints look really awesome in my head..... anyone know where I can get a Neural Downloader? :-P

[Admin]

Hey HDL,

I have a friend at CMU that is working on just such a thing.

Check this out:
http://www.biped.ri.cmu.edu/AMASC/index.html

He argues that pulley driven legs are the optimal leg design for a whole set of reasons. Its worth a look.

[Hal9000]

I'd try to steer well away from that design if I were you to be honest.

Why?

There are 3 parts to each leg and they are all connected with gears. NOTHING I have seen walks like that. Even that new barbie doll horse doesn't do it like that, and I think that's amazing.

My personal idea of mechanic hell.

I'd like to hear why you think i'm being stupid in saying steer away from this design admin! lol

[Hal9000]

http://www.robots-dreams.com/2005/12/not_micromouse_.html

^^ This is how it should be done, if anything, surely.

[Hal9000]

http://isobe.typepad.com/photos/micromouse_2005/mmday1006.html

And all the related pictures

[Admin]

http://www.robots-dreams.com/2005/12/not_micromouse_.html

^^ This is how it should be done, if anything, surely.

Hal9000, I think we are agreeing with each other . . . this link you posted looks exactly like what Robotboy86 is trying to build . . .

[Hal9000]

http://www.robots-dreams.com/2005/12/not_micromouse_.html

^^ This is how it should be done, if anything, surely.

Hal9000, I think we are agreeing with each other . . . this link you posted looks exactly like what Robotboy86 is trying to build . . .

I agree.

Forward meow!

[paulstreats]

Im working on some leg designs that would control motion from the main body rather than having servos on the legs.
I will post some pictures when i have them in cad.

Currently they are mechano style models

[Brandon121233]

The servos should be attached inside the legs on brackets, thus eliminating having to deal with all that gearing
kinda like this

[Robotboy86]

Thanks guys Quiet the explosion of helpful advice. 

I am moving away from geared ideas for the above mentioned reasons.


and $$$

I did indeed look into the idea of "tendons", and have always thought it was a good idea.  I remeber seeing adam on mythbusters with a hand that worked on similiar ideas.  My problem with them is that pulling is very easy, but pushing can present a problem.    The gears were a very bad idea, I have since redesigned a smaller robot that will only have 2 segments(for now) to walk with.  Three servo's per leg though, one on top to rotate, one at the "shoulder" joint to move the whole leg, and a third one to move the elbow joint.(Yes I relize it should be hip/knee lol).  I have some HDPE coming in to build out of it.  I have shrunk the design down to let me use weaker servo's($$$) and also for increased agility/speed.  I will post a new picture of it in a few hours..   

Really appreciate everyones advice

[HDL_CinC_Dragon]

Think of the way muscles work. When you move your arm up, theres 2 muscles working. 1 is pushing, and 1 is pulling.

Thats why you have 2. So one can pull while the other pushes. Obviously whichever one is pulling will be doing all the work but if you implement it correctly like the blueprints in my brain do, you COULD get some pushing power also.


@Admin:
The design he is using is very very different from what my idea is so its still safe in my noggin and not exposed to this harsh world yet I doubt anyone will ever have the same idea as me the way I have it. My idea also allows for multiple digits as opposed to the single digit design hes using.

Hint: Bend your finger. Good. Now try only bending the top part WITHOUT holding the rest with your other hand lol. Cant, can ya?

[Tsukubadaisei]

I'd try to steer well away from that design if I were you to be honest.

Why?

Me too. That design is not good. The reason is very simple: the legs may have 3 motors but only one degree of freedom. In other words, the robot will be able to go forward and backward and that is it. It wont be able to turn around, it wont be able to sidewalk either. If you want a robot that can only go forward and backward then you just need 2 servos in each leg. In order to give more freedom to your robot you need to set the third servo perpendicularly to the other two. Suppose the other 2 turn around the x axis then the third should turn around the y or z axis (preferably z axis like the excellent picture posted by Brandon121233).
Also make sure you have enough torque. 10kg.cm or more(for each servo) is my recommendation for any legged robot.
Admin said something about putting the batteries as low as possible. In this case, I would say put the batteries as centered as possible. That open space betwen those 2 platforms seems to be the ideal place to me.

Also have a look at this link: http://www.lynxmotion.com/images/html/build074.htm

http://www.robots-dreams.com/2005/12/not_micromouse_.html

^^ This is how it should be done, if anything, surely.

Hal9000, I think we are agreeing with each other . . . this link you posted looks exactly like what Robotboy86 is trying to build . . .

No it isn't, look the video carefully and you will see that the robot bends it self in the middle. It looks like there is a servo in the middle of the robot. When the servo turns, the direction of the front legs changes. If you look at the pictures, the rear legs are attached to a platform below the larger platform where the CPU and battery are. And I only know that because I was in that micromouse held in tokyo in 2005 (it was when I built my first robot). 

[Robotboy86]

It could very indeed being able to side walk.  If you look carefully, on top of the platform you will see each leg has a servo.  That would turn the leg in each direction it wanted to, and would enable me to walk. 



So how is this a bad design?  If you look at the servo's on the legs(imagine gears are gone) you can see it could clearly walk backwards.  As such, it would only really 90 degrees of movement per servo on the top for it to maintain 360 degrees of mobility at any given moment.

[Robotboy86]

I have reread your posts and still don't quiet understand...

The three segments were orginally designed for improved flexibility.  You will notice that two segments do not the same level of flexibility as three.  The design it self is solid, the gears were a bad idea and I have since redesigned those.

The rest of it, is good.  Sorry if you disagree.

[Tsukubadaisei]

If you look carefully, on top of the platform you will see each leg has a servo. 

Yes you are right, I am sorry. I didn't see those ones. In that case I don't see any problem. The only thing that was bugging me was the fact of not being able to turn (I have never said it could not go backwards), but I should have paid more attention. And I agree with the flexibility argument, 4 degrees of freedom for each leg is much better than 3. I have built a humanoid last year and I used 4 as well.

[Admin]

Even if Robotboy86 didnt put those 4 servos at the very top to rotate the legs, it could still turn if one side brings both legs together, and the other side walks . . .

[Hal9000]

But then this wouldnt be like a horse yeah? I think you would have stability issues also.

[Hal9000]

Oh wait, this is a four legged robot. Instability wont be a problem. Syke.

[Robotboy86]

so everyone agrees this model is viable?  lol.

I am doing the final modeling tonight, screw holes and such.. then going to print it out at work(big paper printing machines) and cut it out this weekend..  servo's still aren't here, but they are coming.

[Tsukubadaisei]

Even if Robotboy86 didnt put those 4 servos at the very top to rotate the legs, it could still turn if one side brings both legs together, and the other side walks . . .

I see, unnatural but plausible.