Alternative to bearings for tiny robots

Infinity Robotic
4 followers
Quick video talking about rolling contact/compliant joints. These emulate how joints in nature work (two rolling surfaces held together with tension).
#machine #Robotics #robot
Today we're going to take a look at rolling contact joints. I've been working on a robotics project
and just due to the size of the features it's hard to get mechanical ball bearings in here,
and so I decided to use rolling contact joints. And this project is going to be stalled for a
little while just because I've got other things to work on, but I decided to make a quick video
about the joints in particular because they're really interesting and I don't think a lot of
people know about them. So I want to introduce this joint to folks for whatever project you
might be working on. A quick overview is that I'm trying to build an avian or bird inspired
robot. Basically it'll hop around and look like a bird. This was my first attempt and it uses
traditional ball bearings and carbon fiber rods for structural support as well as the shafts going
through the bearings. And these aren't the smallest bearing that you can find but they are getting
close to the smallest. There's kind of just a lower limit as to how small bearings you can get
and that puts some design limitations just based on kind of feature sizes around the outer diameter
and inner diameter of these bearings. So I wasn't overly happy with this design for a lot of reasons
but especially for the bearings. So I started working on a new design which uses these rolling
contact joints. Because the design I'm working on is so small I decided to print out a scaled version
so that we could look at the mechanism a little easier.
Okay so a rolling compliant rolling contact joint is made up of two halves that have some kind of
curved surface and the curved surface on both sides roll against each other. That's basically
it. There's not much more to it than that but there's a lot of nuance into the types of surfaces
you build here and also how you connect the two halves together. So this particular joint is
designed to rotate in this manner and there are two main surfaces that we care about. There's
this rotational surface on this half and the inside section over here. These two outer walls
are just to help provide some rigidity to prevent kind of out of plane movement in this direction
but it's not really the part we're interested in. We care about these two faces that mate and rotate
together. So on this side we've got a flat section that transitions into a gentle curve and then a
steeper curve on the back side before turning into another flat. And then I'll show a cross-section
view of CAD but here we've got a relatively flat section followed by two steep curves on either end.
And that's one of the nice things about these contact joints is that you can engineer the
surfaces that are, you know, mating together and get different behaviors depending on where
they are in the travel which is not something you can easily do with regular ball bearings.
And this is basically how joints in the human body works, in mammals, in birds. You've got
two sides of the joint that are rotating against each other in a non-linear fashion and that
constrains the movements of your limbs. The other really striking feature about these joints is how
they're held together. Typically using a compliant mechanism, either a cable flexure or a solid band.
I'm using fishing line. This is actually called Dyneema. It's a ultra high molecular weight
polyethylene, I think. You can get it in very thin line which is what I needed for
these small mechanisms. But you can use metal cables or metal straps, like whatever works.
And basically you have two loops, one coming on the outside going in and the inside coming out.
And that holds the two halves together. You can see that in my design I have it split up into
four individual channels and this lets me kind of braid or sew the line through the joints and hold
it together which gives you a mechanism that rotates. So I'm going to try to show you how
this is sewn on camera but bear with me it's kind of hard to capture it all. Hopefully it's not too
confusing. But basically we're going to go through a hole in one side up around the channel onto the
other side and then back through a hole. And we just repeat that process making kind of a figure
eight the whole time. So first we'll start on this side. We thread it through a hole, pull it through,
wrap it around the channel, grab our other half and thread it back through the top hole
and pull that through. And that is the basic pattern. So you can see here we've gone on the
underside and then it comes up in the gap between the two and then goes back down into the hole
to the underside. Now what we do is we go to the other side. So you go all the way across
and thread it through the last hole on this side, pull that through, and we repeat the pattern.
So it's from the top now we go underneath through the channel and then we want to come back up
through that hole. Okay now we need to basically do the inverse for the the inner two. So to do that
best way I found is actually to loop it around the outside like that. You can hold on to it
and we'll want to go back through this section here. So going through the second hole on the
underside out the top and repeat the process in reverse. Thread it back through the middle
and into the middle holes on the other half. Thread this back through the middle.
Okay and now it is basically done. So we can tie it off however you want. I found just
looping it through the edge here a few times is sufficient to hold it. Okay so I tidied this up
a bit off camera but you can see that we have our completed joint now. It naturally wants to stop
about this far. There's some compliance there because it's a cable but it doesn't really want
to bend much more than that because there are two kind of steep angles right next to each other and
so that prevents it from wanting to move much further. Whereas this side very easily rolls
forward. Actually it rolls a bit too far forward. This was a design flaw with my current model. It
kind of gets here and then just keeps going which is not ideal. So that's just a mistake on my part
but you can see how the forward versus back rotation motions are very different due to the
inner shape of the contact joints. And if we look closely you can see the pattern. So I've got
on this side it's over, under, under, over, and then the reverse on the top. Under, over, over, under.
You'll notice it's also pretty strong out of plane. So obviously it rotates in this direction
but we don't really want it to rotate in this direction up and down. And there's a little bit
of play there but it's pretty robust and that's a combination of these two outer guards help prevent
it from rotating. And also just this mechanism because it's a cable, the cable doesn't really
want to stretch in a rotational direction. And so applying force like this there's just not a lot of
room for the cable to stretch and pull it in that direction. Whereas this works easily because the
cable isn't stretching at all. It's just shifting where it's landing inside the trench. So it's
zero friction this way and pretty strong in that direction. And this is just tied,
those are just two little knots. So if you actually had a mechanism to hold this cable
firmly in place mechanically this would be a pretty robust joint until that happens. And then
the poor guy has a dislocated joint. So you can see this joint that we built is effectively
this joint here. So why choose rolling contact joints over just a traditional ball bearing?
As I mentioned size can be an important factor. Ball bearings just have a lower limit as to how
small you can make them because they're relatively complicated assemblies. Whereas a rolling contact
joint can be made basically as small as you want. They have extremely low friction. So as you can
see here this is one of my earlier tests. Due to the nature of the two surfaces rolling on each
other and being held together with some kind of compliant flexure mechanism there is very little
in the way of friction between the two halves. And honestly they just look cool which I think
is an important part for robotics. You gotta make them look neat. Now I didn't invent this joint at
all. You can find it in tons of different robotics projects but you don't really see it too much in
amateur or hobby projects which is why I thought you know I'd make this video just to kind of
spread some awareness. Because it is a neat mechanism and it lets you build in my case
very small joints that would be otherwise pretty impractical to do with bearings. If you're watching
this video before June 14th and you're on the west coast or feel like flying out to the west coast
you should come to Opensauce. I'll be there. I think 150 other creators are signed up to be there
and something like 500 exhibits from folks in the community. So I think it'll be a lot of fun. I
really wanted to go last year but it conflicted with something so I just couldn't make it. But
I'm excited to be there this year and see all the cool projects. I'm hoping to bring the little
prototype camera sensor that I made in the last video and set it up with like a microscope or
something so people can take a look at it with their their own eyes. But yeah June 14th, June 16th
hope to see you there. I'll share some more details about this project in the future once I get a
little bit more done. For example you can see there are these cables on the front and back and those
are actually unrelated to the joints. Those are the actuation mechanism and they'll basically act
like muscles that pull different parts of the structure to make it hop. But I don't have nearly
enough of this done to actually share yet so we'll just hang on to that for a future video. But
otherwise I hope this was interesting and just a nice quick primer on a new type of robotics joint.
Thanks for watching and I'll see you next time!
