Experimental Star Wars Speeder Bike

追加: 2026-05-12

[00:00:10]It's time for one ball bike part two.

[00:00:13]You can check out part one to see what's happened so far. Yes, it's a one ball balancing robot that I can ride on.

[00:00:19]Importantly, it's actually balancing on the ball in all directions, which makes it holonomic, so it can move in any direction at any time. So, not just a ball shaped wheel with an axle through it. It went a lot faster than I expected, but there are a few issues.

[00:00:32]One of those is static electricity, which I think is activating my safety system and cutting the motors off. Yes, you can see my hair standing up on end, so it's almost as good as a Van de Graaff generator. I've got a potential solution to that, which we'll look at later. The other issue is that I couldn't stop it spinning around in yaw, which is the vertical axis. I could turn by holding a massive board out to one side to help control it. It's very light to turn on the spot, so the slightest bit of air resistance will cause it to turn, which surprised me. There were loads of suggestions and comments on the last video, so I'm going to try to talk about most of the possible ways we can make it face the way I want or not rotate when we don't want it to. The first possible way to control yaw is with a control moment gyro. I've built quite a few projects that make devices balance using control moment gyros, so here's the recap. When a spinning mass is moved in one axis, it causes a reaction force on a perpendicular axis.

[00:01:25]This is gyroscopic precession, and you can check out my balancing robots playlist for all the videos and more explanation. There's actually a commercial product called the Seakeeper, which does exactly this to reduce roll in boats, actively moving the pitch angle of a spinning mass to cancel out the unwanted roll motion. I built various balancing devices, which measure the angle in roll and then actively control the pitch angle of the spinning mass, which in turn makes a reaction force in roll to hold the device upright. I built a two in line wheel balancing robot and also a one wheel robot, which used control moment gyros to balance sideways and an active wheel to balance back to front just like a Segway. And then I tried to put the gyros inside the wheel, but the results were a little bit weird. So, we could attach some gyros to the one ball bike and actively control them in one axis, so the perpendicular force makes the bike rotate in the yaw axis. That would mean having them mounted vertically with a control axis aligned horizontally, so the resulting perpendicular reaction force is in the yaw axis. The main issues with this are the amount of mass that we'd need to carry, and also that we need to constantly power the spinning mass. Also, the gyros would need to be able to turn around continuously in the control axis, or at some point we wouldn't be able to turn the bike any further, which means powering them with slip rings. Another suggestion was a reaction wheel, which is generally a spinning mass, but not to be confused with a gyroscope. A reaction wheel is fixed in position, but it varies its velocity in either direction around a stationary center point, so the reaction force causes the device to move in the opposite direction. In my previous projects, I used a reaction wheel to balance a robot upright, but we could simply place a large reaction wheel around the ball of the bike, for instance, and use the reaction force to push the bike around in yaw as the wheel turns in the opposite direction. This would probably require quite a bit of mass, but also the reaction force only occurs when the reaction force is accelerating or decelerating. So, at some point it would be going as fast as it could, and then we wouldn't be able to turn anymore without letting it slow down and accelerate again, or turn the other way with a rapid deceleration of the wheel. And more importantly, if the reaction wheel is spinning and the bike tips in roll or pitch, which it always does to move, we'd still experience the force from the wheel's accidental gyroscopic effects, which would make the bike tip in a perpendicular axis. So, all together, I prefer something simpler, which I don't have to compensate for. Quite a lot of people suggested just rotating the ball, and that would be easy by turning the back wheel around 90°, so it's active horizontally instead of vertically.

[00:03:54]Currently, only two wheels are active when traveling sideways or balancing in the roll axis, and the three wheels are active when traveling forwards and backwards or balancing in the pitch axis. So, turning the wheel around would just mean that the back wheel would now be active when traveling sideways in the roll axis instead, but we'd also be able to rotate the ball in yaw. I'm not totally convinced that we'd get much traction though, because there's a very small contact point between the ball and the ground, and it's so light to turn that it would be hard to control, both to start spinning in yaw and to stop spinning in yaw. I'd like to try this though, because I've never seen a ball balancing robot with two wheels vertical and one horizontal, but I'd like to do it as a smaller project before I rebuild large portions of the bike, so look out for that in the future. So, what's the answer?

[00:04:39]Yes, it's fans. We're going to blow myself around with these electric ducted fans I'm going to place somewhere to blow me round to make me rotate. Now, these I've used before in the single caster jet skate project that you should watch, and that was enough to blow me along, so it should be more than enough to blow the bike around, because basically it would turn if I just held out a board to one side because of the air resistance, and it will continue to blow me round. So, we haven't got any problems with reaction wheels or gyroscopes saturating and not be able to turn any further. This should do it just fine. But, we just need to find somewhere to mount them and some way of controlling them. So, I 3D printed the EDF mounts, and this is all Polymaker PolyMax PLA. Thanks to Polymaker for supporting the channel with filaments, and I've got some pins that go through some holes, and those are placed to hold the EDFs in through the little holes in the ears, so they don't blow themselves out. And of course, there's two of those. This is my frame, which is an extension of the front to hold everything, and this is just 2040 extrusion, and then I've got some bumpers, which I really needed.

[00:05:43]So, I've got the bumpers on the front, hopefully we don't need this time, and I've got my fans mounted, and you'll notice what I've done is the intakes, where this bell duct is, are on the inside, so air gets sucked through here and blown out, and that's inconveniently why the wires are on the outside, but that's the way they make them. I actually don't know if it makes any difference if I had them the other way round, but it feels like blowing air onto something on such a short distance wouldn't be so efficient. The other reason they're that way is because these are a bit of a finger chopper. They're very powerful, and if you get your hand anywhere near it will suck your fingers in. They've got metal blades that will slice the end of your finger off. So, really don't want those on the outside, where I might fall on it as it's running. At least air's being blown out.

[00:06:24]Anyway, let's mount this up and see how we're going to power it and control it.

[00:06:27]So, here's the new overall look of the bike with those pieces attached to the front, EDFs and the bumpers, and we've also got space for a battery there to power the EDFs. And that's because they run on a separate voltage with 200 amp ESCs. I've got some power distribution with a switch and a little power meter, so we can monitor that battery.

[00:06:47]So, that's power, but we need to give them a signal, which is just servo PWM into the ESCs to control those EDFs.

[00:06:54]I've got a servo tester here for now, so we can power one of these up.

[00:07:00]They're mildly terrifying.

[00:07:05]So, yeah, we're going to have no problem spinning the bike round, and I think we we don't need anywhere near that amount of power. So, it's actually a much better balanced bike now, because we've got this mass on the front, and it makes a surprising difference, so I can sit back quite a bit without going backwards. My knees aren't resting on the axle ends anymore, which is a bonus, although I'll still probably put some knee pads on.

[00:07:26]So, that's actually much more comfortable. That's a surprising difference that extra mass makes. So, that's good. But, now we need to control these EDFs, so I can spin round. So, as you can see, it's actually very light to turn, and it keeps going with the slightest push.

[00:07:41]So, we don't need their full power at all.

[00:07:44]But, we do need some way of controlling it, so I was thinking about some sort of foot pedals, but I think I really want to keep my feet still. It's going to be quite hard to operate pedals on there, and my feet are nearly touching the ground anyway, so that's quite uncomfortable. So, what I think I want is some sort of hand control, and we've already got those twist grips, but they are already in use to bias the set points for the pitch and roll balancing, and that's how I was able to throw the bike around so much at the end of the last video, before I crashed at least. So, let's have a think about that.

[00:08:20]Are you really using those twist grips now?

[00:08:24]Push it.

[00:08:26]God, bloody hell, this is well.

[00:08:34]Oh, yeah, they're really good for braking actually.

[00:08:40]Have you ever thought about getting CNC metal parts made like these metal plates that I used in this project, or maybe resin or SLS parts, but you don't have a suitable printer?

[00:08:49]PCBWay is a reliable one-stop shop for PCB manufacturing assembly, but also other types of manufacturing services, including contract manufacturing, all under the same roof.

[00:09:00]PCBWay manufactures all sorts of printed circuit boards, but PCBWay don't just make PCBs, they also provide parts manufacturing services, including online CNC machining, sheet metal fabrication, 3D printing, and injection molding.

[00:09:13]PCBWay's 3D printing services include SLS, SLA, DLP, FDM, in a variety of materials, including various resins, nylon, PLA, ABS, TPU, PEEK, and more.

[00:09:25]There are various surface finishes available, including painting or even electroplating.

[00:09:30]PCBWay have an open source community called Shared Projects, where many people have uploaded their project source, so you can build the projects, too. If there's a part like the PCB that PCBWay can manufacture, you can buy this directly from the project page, otherwise the source files are included as downloads. PCBWay also now have events, and it's time for the KiCad PCB design competition to celebrate the release of KiCad 10, which was launched this year. Top prize is $500 cash and a Raspberry Pi 5, but there's also a participation award. Use my special coupon code at PCBWay.com, which I'll put in the video description, and you can get $10 off on orders over $30 on PCBs and assembly, CNC, 3D printing, and other services. I think I really want to use the left-hand twist grip to control the EDFs so we can turn in the yaw axis, and this would make the controls the same as the two-ball bike, which could turn on the spot. At the moment, that twist grip makes the bike lean over sideways in the roll axis more, but we probably need that functionality as well to make it lean into the turn. So, my plan is to put a kind of shift switch on the left-hand twist grip so that I can either control just the EDF or I can offset the set point in the roll axis as well.

[00:10:43]I've attached that switch to my left-hand twist grip, so now I can rest my weight on the twist grip, and I can press it if I need to.

[00:10:51]So, let's try one of those EDFs. It doesn't take much to get me spinning.

[00:10:55]You can see it's blowing air everywhere, and once I'm spinning, then it's pretty easy to keep going, so I don't need to keep thrashing it, and I can stop by turning the twist grip the other way and firing the other EDF, which means I can stop fairly reliably, so I've got quite good control authority.

[00:11:14]Right, let's try and drive in a circle.

[00:11:16]I've only got limited space here, and it's quite a lot of concentration to balance on this anyway, both sideways and back and forth. So, what I'm doing is powering one of those EDFs and trying to lean forwards at the same time so that I can drive forwards and turn left and point in the direction that I want to, essentially, without randomly spinning rounds. That actually seems to work pretty well. We could do with some more space for testing, though.

[00:11:52]Of course, if I had the EDFs further from the center of the ball, further in front of me, then I'd have better leverage and better control authority, but the bike is fairly compact at the moment, which is easy to transport and handle. Let's talk about all the times I fell off the bike in part one. I'm pretty sure I accidentally hit the e-stop the first time, but the other times were pretty random. It looks like there's a lot of static electricity, which is causing some glitches, and I'm pretty sure it's hitting the IMU, which measures the angle, or at least the I2C bus that the IMU is connected to the Teensy with. I implemented a safety feature in my code, which cuts all the motors off if the bike tips over 20° in any direction, and that means I don't fall into the wheels spinning under full power. But, it checks the angle 100 times a second, so any glitches in the data will cause it to trigger immediately. So, I've used an additional timer to check the angle has exceeded 20° for more than 1 second. So, this should stop it triggering if there's a glitch in the data. But, what's causing static in the first place? If I had to guess, I'd say it's this massive polypropylene ball running on rubber rollers, which sounds a bit like the description of a Van de Graaff generator. It's the sort of thing you build if you wanted to generate static.

[00:12:59]Obviously, rubbing plastic is going to cause static. Pretty sure that's where it's coming from. Had these problems with the two-ball bike as well, but not in the screw bike, which worked fine with the same controller, but no big plastic balls. Lots of people suggested brushes that conductive to go and ground the static or drag chains and lots of things, but what about if we just make the ball metal?

[00:13:21]So, I've got this ball, which is polished stainless steel. Now, it looks pretty round. It was intended as a garden ornament, and I got it off AliExpress for not very much money. So, we'll see how it goes. Now, it's only half a millimeter thick steel. This is 10 mm of plastic, so I'm not sure if it's actually going to support the weight of the bike and me or whether it will just crumble and turn into like shredded sharp steel. So, I haven't tried it yet. We'll try it when we do a proper test.

[00:13:50]Right, let's try and do a bigger circle in here.

[00:13:54]He says, "Why are we going this way?"

[00:14:00]Well, it's turning. Oh, yeah, here we go.

[00:14:06]>> [music] >> And once it's spinning, it keeps [music] going, but actually it's not too bad on this floor. Right, let's just try and um I'm going to drive rounds anticlockwise now, hopefully.

[00:14:19]Come on. Come on.

[00:14:24]Oh, yeah. Well, it's easier if I'm rolling and the fans flat out.

[00:14:34]Once I'm rolling, I don't need so much fan to keep going, so still turning. Come on, forwards a bit more.

[00:15:05]Well, let's see if I can almost [music] do a figure of eight. Oh, no, that's bad. Whoa!

[00:15:10]Whoa!

[00:15:12]Um That's much worse than last time, even though Well, it didn't cut out.

[00:15:18]I don't know if these fans are causing me more problems.

[00:15:21]Huh. Right, let's just try one more.

[00:15:24]That was a bit scary.

[00:15:26]I'm going to try and turn the other way.

[00:15:39]Well, it works as intended. Oh.

[00:15:41]But, every time I do it, I get a little jolt, which is probably [music] static, and it's much worse than it was.

[00:15:47]Did someone say crash helmets?

[00:15:49]This one?

[00:15:51]Hmm, maybe something more substantial.

[00:15:55]It's good luck if you break mirrors, right?

[00:16:00]Yeah, this is the one.

[00:16:08]Right. It's nothing like Star Wars, actually.

[00:16:18]>> [music] >> I think this fan's made it worse.

[00:16:56]I'm not sure why.

[00:17:01]Oh, well, it's the end of that silliness.

[00:17:04]So, I think the static electricity definitely is coming from the big plastic ball, cuz when I touch it, I get static shocks. So, lots of people suggested ground chains or some sort of grounding method, but what if we use this metal ball? So, this is stainless steel. I got it off AliExpress. It was 200 quid, and it's a garden ornament.

[00:17:22]Um let's see. I'm not sure how round it is.

[00:17:25]It does seem to roll okay.

[00:17:27]I can see a weld line round it, but it's not really wobbly.

[00:17:32]So, it is only half a millimeter thick, though, the shell, which means it might just crumble into a mess of broken metal when I sit on it. But, let's see what happens.

[00:17:44]Oh.

[00:17:49]Oh.

[00:17:51]Why are the wheels touching the ball?

[00:17:53]Oh, the bike's touching the ball.

[00:17:55]That's loose. Why is it loose? It's supposed to be the same size. Um right, I think that's too small. It's supposed to be 600 mm as well.

[00:18:03]Is that Does that look smaller?

[00:18:05]It's slightly smaller, actually.

[00:18:07]Ah.

[00:18:08]There we go. That's AliExpress for you.

[00:18:10]600 mm is about 550.

[00:18:14]Right, I was going to try and move all the wheels in to accommodate the smaller ball, but actually I can't because under here is the end of the extrusion, and there's a cap in the plastic bracket, so I can't slide it in that way. And I can't get them closer anyway cuz of the battery holder.

[00:18:28]I don't know if I can move the back one in, and if I did, then the only wheels wouldn't all run smooth on the ball anyway, so it's all been designed for a bigger ball. So, I think what's happening is static is probably affecting the inertial measurement unit, and that's the bit that measures the angle, cuz it's giving me all sorts of spurious data and glitches. It's either that or the I2C bus into the Teensy, but there's not much I can do about that now. I'm going to have to do some testing, possibly even alternative device. I already nickel screening sprayed the whole box, and it was much better than the two-ball bike, but uh now it's really bad, and it's a much drier day today than the last time I shot this. So, I think probably the dry heat is affecting this, and I'm definitely getting static shocks off the ball, which I wasn't last time.

[00:19:10]Come on. Come on. Turn.

[00:19:14]Okay, so the plan vaguely works.

[00:19:17]We're going to need a proper plan to make this work properly, either a proper metal ball that's the right size, that I'll have to have custom made, or we're going to need to shield static from the IMU or find an interface that isn't susceptible to massive static discharges. But, that's something I'll have to investigate.

[00:19:33]>> [music] >> It's still a bit uncontrollable. I mean, it's still turning, and I have to compensate, but I'm just about in control.

[00:19:51]Whoa!

[00:19:53]Oh.

[00:19:54]Felt the static in my hair then as well.