help with 3 DOF joint for robotic arm

[guncha]

Hi!

I need some experienced advice from you guys. I've been working on this idea I had couple of days ago about this robot that has tentacle legs/arms (a la Matrix), but I'm stuck because of this minor detail I overlooked. It's quite hard to describe in words what actually is the problem, so let me just show you a picture:


(it's not finished)

So that's a joint for the tentacle leg, that has three degrees of freedom (correct me if I'm wrong) and can move around for something close to 180 degrees. All is well in theory. That sphere in the center was supposed to be a ball joint, but today while looking for parts I found out that ball joints can't do full 180 (yeah, what was I thinking), only something like 65 degrees. And I just don't see any other way how to make it work. What are your thoughts on this?


Sooooooo every comment, point or idea is welcome.


P.S. Btw, this is a great site and everything. Especially the jokes about kittens

[hazzer123]

Can the joint rotate in the y-axis? (i.e. the axis pointing in the same direction as the arm?) It must be able to in order to be 3DoF.

The picture you showed us covers up some of the important stuff in the middle of the arm where the servos interact with the ball. You should make a picture from another camera angle.

[guncha]

Okay, so I guess it's not 3 DOF after all. 

The thing is, neither servo interacts with the ball. At least directly. Would be sweet if they could, but I don't think there is a way. They do it with the help of those top levers, which in turn are driven by multiple belt drives and shafts. Now, after reading some info on gear chains, I realize that belt drives wasn't the best choice.

[Admin]

So that's a joint for the tentacle leg, that has three degrees of freedom (correct me if I'm wrong) and can move around for something close to 180 degrees. All is well in theory. That sphere in the center was supposed to be a ball joint, but today while looking for parts I found out that ball joints can't do full 180 (yeah, what was I thinking), only something like 65 degrees. And I just don't see any other way how to make it work. What are your thoughts on this?

Not true, my shoulder/arm ball in socket joint can move like 290+ degrees. It can do this because there is flesh holding it in place so that it doesn't fall out. Just find a flexible way to hold the ball in.

Or perhaps think out of the box and take it from an entirely different approach. Perhaps a 3D 'four bar linkage'?

[guncha]

Not true, my shoulder/arm ball in socket joint can move like 290+ degrees. It can do this because there is flesh holding it in place so that it doesn't fall out. Just find a flexible way to hold the ball in.

OMG! Mine too!! Yeah, I totally forgot to check with the worlds best designer. I figured that the best way to keep the joint-ball in place would be to use two cables that are attached to the ball and run around the thing that holds it freely. That way it would be fixed on all axes, but still able to rotate.

Thanks for the idea. This thing is coming real nice.

[paulstreats]

For advanced reference, new scientist wrote an article about a robot octopus tentacle that is being developed http://www.newscientist.com/article/mg20127006.500-robot-octopus-will-go-where-no-sub-has-gone-before.html

basically its electrocative segments that expand or contract due to electrical imulses causing a system similar to realistic muscular tentacles.

You idea is good but realistically you might only be able to fit 10 segments together for a metre long unit whereas the above system could have 100+ segments per metre causing a far more manoueverable tentacle.

Also have you looked at weight consideration? after you connect 5 of these segments together than will the first servos be able to hold the rest of the tentacle? I mean that if each segment has 3 servos than that is equal to 4 x 3 servos plus 1 servo supporting the weight of its corresponding 2. meaning that fo 5 segments, 1 servo will have to bear the weight of 14 servo's before you even look at other materials....

I hope the link above leads you to a better design

[Admin]

For advanced reference, new scientist wrote an article about a robot octopus tentacle that is being developed http://www.newscientist.com/article/mg20127006.500-robot-octopus-will-go-where-no-sub-has-gone-before.html

basically its electrocative segments that expand or contract due to electrical imulses causing a system similar to realistic muscular tentacles.

I hope the link above leads you to a better design

Well actually they only plan to use artificial muscles, they don't actually use them. I can say for a fact that the worlds most advanced artificial muscles in the labs today can only supply about 1% the force needed for that octopus, and are so inefficient they are useless outside the lab. I know this as my coworkers work on artificial muscles for the Navy, and I've done the calculations for them for robotics. Probably another ~20 years before the technology matures enough . . .

That being said, I think flexible parts/joints for robots is the future. My robot fish moves using flexible pectoral fins, and would never work nearly as well if it was rigid. But its much harder to model and analyze flexible parts - think about the kinematics for an infinite number of degrees of freedom flexible tentacle, and you'll understand why.

[guncha]

Also have you looked at weight consideration? after you connect 5 of these segments together than will the first servos be able to hold the rest of the tentacle? I mean that if each segment has 3 servos than that is equal to 4 x 3 servos plus 1 servo supporting the weight of its corresponding 2. meaning that fo 5 segments, 1 servo will have to bear the weight of 14 servo's before you even look at other materials....

I hope the link above leads you to a better design

Ah, yes, lightweight and soft artificial muscles that could contract more than 30% in reasonable time would be mighty super sweet (smiley with a dream bubble overhead)

And that's a very good point you made, paulstreats. I did indeed look at the weight consideration and things that follow it and that servos are no way to go (I didn't like them from the start). There's just too much extra weight to swing around, meaning very strong strong servos meaning couple of thousand dollars only for a single leg/arm. Btw, I'm thinking three 2 DOF joints for a tentacle would be cool enough.

Now back to the weight problem, lo and behold, the fruit of my sleepless nights:


Not sure what to call it, but the principle is very much like in Maglev trains (the ones that fly over magnetic rail and use magnetic propulsion to move ahead). Sort of a linear stepper motor. Neodymium magnets being strong and cheap that they are could be quiet and lightweight replacement for using dc motors. I am still yet to figure out the right size and strength of magnets, a compromise between holding torque and step size, but I'm already convinced that this is the way to go. A big pain to actually calculate something because of the ridiculously complicated way magnetic forces work, so I'll just have to experiment and see what works and what not. Waiting for the money to order parts. I'll keep you posted.

Geez, I seem to have verbal diarrhea. That's it, time to sleep. Bye!

[madchimp]

not sure you wont have the same weight issue  with magnets but so far the cheapest place I've found on the web for rare earth magnets is magnet4less.com just thought i would throw that out there took me awhile to find that site