Alternative to bearings for tiny robots

Cool stuff. I work a lot with flexures, didn't expect you to make a video on a compliant mechanism. These kinds of joints are indeed quite common in low friction robotics and mechanisms (such as tele-operation with haptic feedback). Your joint has very low sideways support stiffness, as the compliant elements you used are essentially wire flexures (at least as long as you keep the wire in tension). They are compliant in the bending mode about the knee axis of rotation (which you obviously want), but also a similar bending mode in the rotation axis perpendicular to the knee rotation axis and perpendicular to the length of the constraint. Pure rotations in your design are somewhat compensated for due to the spacing of the wires, but pure translations sideways I expect the joint to have nearly zero support stiffness as well. If you want to improve this, use blade flexures of some kind. You can perhaps just use thin metal sheets to fabricate these out of, I believe they are common in hobby stores for scale models. Considering the scale of the joint, the metal sheet you need to use is likely so thin you could cut it with a razor blade. Alternatively you could add a supporting flexure in parallel with the joint you already have (for example a folded leaf), but to do that properly is likely too hard without experience in flexural design. If you want, I can give you a book recommendation that would significantly help you with such joint design (and mechanical constraint design in general).

Finally robots can also have bad knees.

I've seen rolling contact joints used for telescope pointing and some 3d-printed flexures, but never considered non-circular profiles. Can't wait to see the teeny-tiny birdbot in action!

These are also considerably lighter than metal ball bearings. The main problem with these is strength. Biology has incredible designs for ligaments, bones, etc that are very difficult to match with traditional materials. Fun stuff!

Suddenly dislocating joints make a lot more sense seeing this. I can see this rolling geometry allowing for some cool variable strength grippers

I love how in picking up the rolling-contact leg it instantly is recognizable as something more biological and light, as opposed to the bigger clunkier mechanical bearing-based design These rolling contact joints are something I became interested in recently while trying to find a compliant but weight-bearing hinge, and came across a paper from the 2000s where they used it to make satellite solar panel deployment mechanisms

Something to think about if you are not already would be to use nylon for the joints, nylon is self lubricating so it should reduce friction. It may not have a huge impact as friction is already low but it may be worthwhile to test.

Did we just see the start of Mecha-Dabchick ?

I first learned of this joint as a kid, with the Rubik's Magic puzzle. That thing fascinated me, not for the puzzle, but for the compound folding joints it used.

That rolling contact point joint version looks eerily like a real leg. I was listening to what you were saying, but my brain immediately went "oh, thats a bird leg"

10:08 That's how my last digit of my ring finger dangles if I stretch my other fingers and contract the ring finger entirely. Try it. Probably works because it's the same type of mechanism lol. 9:27 And I felt that in my knee.

I am impressed that not only have you recapitulated the human knee joint, but also the cruciate ligaments which support it.

2:14 That undesired movement is called "Side Fumbling", and what you have created there are some rudimentary marzlevanes.

I've been rolling some joints myself and these are really cool!

I too am fascinated by rolling joints 🔥

It's like these glasses cases! With the three stripes, if you know what I'm talking about. Edit: I think they're called "magic cases". They were all the rage back in the late 00's

I usually do a makeshift rolling contact joint to explain why training the opposite of a muscle is just as important. Also helps with teaching correct deadlift form and anterior pelvic tilt.

Any references on how to design the curves in the joints?

I started my version of this back in 05. Here's what I learned. You can make them cams to increase torque advantage to position ratio. The "ligament" cables tent to abrate on each other and the groves. I was able to mitigate this some by Drafting the groves by 5°. That's as far as I got. Good luck.

I've seen those with stainless steel bands, they provide a bit of stability/stiffness to the joint as well. Nice design :)