YouTuber Builds Onewheel With Tracks Instead And It’s Not Great

The one-wheel is a triumph of modern sensor and control technology. That made it possible to sense the acceleration and position of a platform with a single wheel, and to control that single wheel to keep the platform stable and level, even in motion. [RCLifeOn] has now taken that same concept and made it more hilarious by swapping out the wheel for a track.

The original idea was to build an electric snowboard, which worked just okay. Then, it morphed into a tank-based one-wheel instead. It’s a bit silly on the face of it, because a track is more stable than a wheel. That’s because instead of balancing on a small flattened spot of a tire, it’s got a wider, flatter footprint. But that means there’s no real need for balancing control as the track is statically stable.

The 3D-printed track assembly is driven by a powerful brushless motor via a gear drive for additional torque. Riding it is difficult on 48-volt power as it easily throws [RCLifeOn] off the board with its raw torque. At 24 volts, however, it was just barely ridable with some practice. But it was ultimately pretty terrible. It was either not moving at all, or jerking so hard that it was impossible to stay on the thing.

We’d like to see this concept tried again, perhaps with a rubber track and a more refined controller. Video after the break.

15 thoughts on “YouTuber Builds Onewheel With Tracks Instead And It’s Not Great

      1. Graph x=y and y=log(x) for a good idea of the difference. For some purposes, such as audio volume via resistance, logarithmic pots are experienced by our ears as sounding smooth, while linear pots will give us a sudden and sharp increase or decrease.

          1. Because he can’t get started. At the moment, when he turns the contraption on, it receives a bolt of juice and it runs away without giving the “pilot” enough time to adjust his balance. He needs a smooth acceleration in the low speed range, so he can adjust his balance to the acceleration forces.

          2. Imagine if you had a 500 HP car. It takes only 20-30 HP to drive at a decent speed, so for normal driving you only need 5% of the engine power. Now imagine if the throttle pedal response was linear: you would press it in 5% and you already got enough power to go 65 mph.

            For a car, the saving grace is the two tons of metal you need to accelerate, so it won’t instantly fly off the handle. For the motorized board, it’s just instantly going to flip you over with that much excess power. If you reduce the maximum power, like the guy did by removing batteries, then you have to press the throttle all the way in to get going and your throttle response will just be on/off – again, very difficult to drive.

            With logarithmic throttle response, you have fine control for taking off. When you speed up, your wind resistance is going to go up non-linearly (V^3), which is similar to the logarithmic (or exponential) curve of the potentiometer, so your throttle response at higher speeds becomes more linear and again you get finer, more predictable control of your speed.

  1. I think the treads might be too flat in profile and too sharp. You’d want them to be taller in the middle and round off towards the sides (to steer smoothly when you lean), and less sharp so they don’t sink into the ground and reduce this difference in circumference.

    If you need more grip on ice, maybe heat up some nails or tacks and sink them into the tread plastic.

    As somebody else alluded to, smoothness of motor control will be crucial. This is because treaded vehicles decelerate INSTANTLY. Not just quickly, but immediately. The average ground patch for a car tire is the size of your palm, for a bike it’s maybe a few square cm. The ground patch for a tank is the size of a studio apartment. Even with all the weight of that armor, it will go from highway speed to zero at the blink of an eye and God help anybody tailgating (or riding on top, as is your use case—balancing under such deceleration is very difficult). Might want to mess with the firmware to apply a gradation to the brakes (or reduction of motor power) so it can’t hit you all at once. Since there is little wheel rotational momentum and power is basically always needing to be applied to maintain speed, you need to have the speed ramp up and down at a rate that your reflexes can accommodate. Potentially this is just a software fix.

    1. And of course the need for balancing by varying the speed of the wheel is reduced by the treads being much flatter and larger than a wheel, so you can get away with far less throttle responsiveness than you would need for a regular one-wheel vehicle

  2. Or how about this? If it’s on treads, maybe it doesn’t need to be self-balancing. Maybe just add a little more tread and you can just stand on it. Seems like the control system is not his friend.

    1. How about this?
      It’s just a bad idea.
      Like a split rear wheel bike, a table made of an old tire and twine, mySQL, server side Javascript, women wearing shoes etc etc etc.
      Not even a pulsejet can save them!

      Unless the whole point was clicks.
      I don’t know how much hackaday charges for posting a youtube clickbait link article. It’s too much.
      Does anybody bite anymore? Don’t click that vid.

  3. Interesting, referring to a mechanism like this as a “onewheeler” as it clearly consists and rides on two wheels? The fact that the wheels are hiding underneath some tracks doesn’t make it a one wheeler, does it?
    A onetracker perhaps?

  4. This is an example of why, if you are trying to improve on an already existing design, that you should try to retain as much of the other features of the original as possible. For every aspect you ignore, you are infinitely increasing the scope of your project. I think he would have succeeded if he retained more of the “one wheel” concept in his design.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.