brainless unicycling...

[YooNeeNoob]

...and here's me thinking we were doing something special.

Unicycle - YouTube

[cristianMN]

wow, very interesting
did you make it ? do you have instruction about making it ?

[Nurse Ben]

Can't wait to see the muni version

[Chrashing]

Elegant.

It seems obvious how it stays balanced forward and backwards, but side to side, I'm not sure. Is it something to do with the torque of accelerating then slowing the flywheel? Or is the flywheel unbalanced and the weight moved from side to side?

How long does it stay balanced on a single charge? mins, hours, days?

[Mikefule]

It looks like it's the reaction to the flywheel - whereas a "real" unicycle keeps upright in the "side to side" direction by a slight "weave" in the track followed by the wheel.

[aracer]

Very nice. I have a 3d accelerometer module for my RPi here - now I know what to do with it!

[Onions]

I got into unicycling through an electronics forum, where Mowcius was talking about unicycles and it caught my interest. On the same forum there used to be a lot about self-balancing robots like that, only most of them had two wheels. They are (theoretically) simple to make, but very interesting things (or I think so at least ). An example of one can be seen here: http://arduino.cc/forum/index.php/topic,119098.0.html; there's plenty of others too.

Very nice. I have a 3d accelerometer module for my RPi here - now I know what to do with it!

I'd have a look at the Arduino boards (see above link); a RasPi is overkill for it, imo

[aracer]

Originally Posted by Onions

a RasPi is overkill for it, imo

Well of course it is - but that's part of the point (though I do also have an AtMega board and some discrete AtMega processors which I program at a lower level than Arduino I could use if necessary).

Though on further thought, I probably need gyros rather than accelerometers for this - a bit more research is needed.

[aracer]

Originally Posted by Mikefule

It looks like it's the reaction to the flywheel - whereas a "real" unicycle keeps upright in the "side to side" direction by a slight "weave" in the track followed by the wheel.

It's a good way to build a robot as you're just controlling a single axis with the flywheel making it less complicated - and don't need to actually move the wheel at all. Effectively that robot is doing a still-stand - videos I've seen of people doing still-stands have them doing a similar flywheel thing with their arms.

Though as a unicyclist, I want a robot which works like somebody riding a unicycle - ie if it's not moving along it should balance by idling. I've had a think about this and realised that all you need is a joint to rotate the top portion of the robot which would allow you to steer the wheel in the same way as twisting your shoulders.

Originally Posted by Onions

I got into unicycling through an electronics forum, where Mowcius was talking about unicycles and it caught my interest. On the same forum there used to be a lot about self-balancing robots like that

So the question is, has anybody ever made a unicycle robot like this? I find I can get a 3-axis gyro module for £4 so it seems worth having a play, and issues with driving motors are solved by using servos for all moving parts (plenty of movement to idle even using an unhacked servo). I'm not totally sure I'm up to writing the control algorithms for this, but there is presumably plenty of stuff out there I can copy. To get the wheel to idle might only involve increasing the gain in the feedback loop to get oscillation.

What would be really cool would be to build a robot which learns how to ride by itself, but whilst I did do some AI stuff involving learning algorithms as part of my degree (and I've even been involved in developing software with learning algorithms fairly recently - though in that case somebody else did all the maths), I'm sure that is well beyond me!