this is why spacecraft that use control moment gyros for stabilization always have some other way of holding attitude. if the spacecraft is subject to some unbalanced torque (from radiation pressure, magnetic fields, etc) the CMGs will move as far as they can go and become saturated. to unsaturate them, they're moved back to the neutral position and the spacecraft's thrusters are used to counteract the resulting torque. the term for this is momentum dumping. the robot crashes because it has no way of doing this. being tilted over results in a constant torque which will quickly begin to saturate the gyros, so that they will be moved very far in one direction even when the robot returns to vertical. unless you plan on doing the exact same amount of left turns as right, this concept fundamentally cannot work. now, this *does* give me the idea of creating some kind of robot that uses CMGs for most of its control, but has compressed air jets for momentum dumping.
The robot could unsaturate the reaction wheels by leaning a small amount the opposite way and letting gravity unsaturate the wheels (I think, not entirely sure). But yeah, it has to be done standing still if you don't want to turn.
@@huvudpersson7344 pretty sure this will work. You just need a bit gyro power left to flip the leaning side and let it hover there until desaturated. would be fun to implement this in the stabilize algorithm and correct its saturation levels as it goes. To solve the forward backward leaning problem of the gyro add them to a slide that corrects its position back to center. Could even be used to help control steering. I would love to work on the algorithms doing all that in parallel. Must be fun to watch afterwards doing its things to balance.
I would love a pogo-stick type platform that could carry a person. It can handle stairs and irregular terrain, so an alternative to a wheel chair or stroller.
Most people don't know how bicycles actually work, to turn left, you first have to turn right, to change your cog to allow you to then turn left, Veritasium did a great video on this. Seems like your monowheel is experiencing this exact same mechanic.
nope you dont have to turn at all. dont repeat information if you never actually did it yourself. countersteering is useful for making sharp turns but is not how bicycles always turn
If you're riding a motorcycle at speed, you must point the wheel to the left to turn right. Pointing left causes the mass to lean right, this turning right. The wheels are the gyros. Bicycles and motorcycle actually respond the same. If you're moving parking lot speed, left is left, and right is right. But as your speed increases, the more influence from the gyro, inverting your correlation. If you get your bicycle up to 30mph+, you will experience this.
You need a slower PID that takes the gyro position as input and changes the left right angle setpoint as it's output. That way it will lean itself left and right automatically to keep the gyros from hitting their end stops.
Actually PID is not a good control strategy. The force vector being so close to the center of mass causes a bifurcation in the systems phase portrait. The system is not guaranteed to have unique control solutions. So the PID assumes solution uniqueness, i doubt PID is sufficient. lowering the center of mass more will make the pid work better
13:03 did you see Veritasium’s recent video about steering? I wonder if you need to apply the same principle. In his video he shows how a bicycle has to steer the WRONG way first (just a little bit and very quickly) before steering the RIGHT way. It’s fascinating and might be helpful!
Nice connection, although it doesn’t really apply here since you reverse steer to get the bike tilted in the direction you want to steer, and here you use the gyro to control the tilt. Imagine if you added a gyro to a bike, then you would never need to reverse steer since the gyro takes the role away from the front wheel. Actually you wouldn’t need to steer the front wheel at all but I imagine that would be a very difficult bike to ride.
You are right about the gyros, assuming they interface with the bodies containing them with torque around every axis. This is however not the case since the gyros here spin in the "same" plane as the wheels. They regulate pitch in that manor only. It is only when you offset the planes the gyros rotate in that the resulting forces are exerted push side to side. In the bike case the force is applied below the center of mass causing the bike to tilt towards the direction you want to turn. (The bike rides out from under you initially) If this force is applied below the center of mass it is likely similar to the bike but less severe. Take note that the two monowheels regulate pitch differently because of their location with respect to the center of mass. This supports the idea that it is indeed what vertasium describes.
@@molomono9481 >This is however not the case since the gyros here spin the the “same” plane as the wheels. The way the gyro is used doesn’t matter in my explanation, as a reaction wheel or control moment, as long as it controls the tilt left and right. >If this force is applied below the center of mass it is likely similar to the bike but less severe. For gyros their position relative to the center of mass doesn’t matter, just the orientation, which is counter intuitive. Although you are right that the unwanted steering is somewhat related to the center of mass. It tries to steer the wrong way because the wheel is pitching down while turning (center of mass at top), which tilts the roll axis downward so when the gyro tries rolling right (to offset gravity when turning left) it spins the wheel to the right at the same time. If the wheel was pitching up to drive forward (center of mass on bottom), however, then the reverse would happen and the axis tilt would actually help with the turn. So the center of mass affects which way the wheel pitches to go forward, which affects how the roll axis is tilted, which affects turning.
@@corporalwill123 haha yeah we agree on the steering but are focussing on different interfaces of force in the system to explain it. Would be interesting to play with both variables and study what is really going on Its weird though since you would expect that the gyro exerting force normal to the center of mass in xy plane would be simple since it is removing non-linearity but it is actually near a bifurcating region making linearized control have non unique control solutions. In other words, PID wont work well
Random project idea: make Gizmoduck from the show “Ducktales.” I loved that cartoon as a kid and ever since I discovered your channel and the possibilities of 3D printed self-balancing robots I’ve want to see a real robot duck riding around on a single wheel. As always thanks for your great content.
I think tire shape might be a sneakily important factor. A flat-bottomed tire might seem easier to stabilize when the wheel is stationary and near vertical, but once you tip over the edge, the gyros are bound to struggle at correcting the sudden change in forces. Plus, when the wheel is moving, it's gonna be hard to make minor steering adjustments (like the video mentioned) since the gyros' smaller adjustments don't really lean the wheel at all. With a rounder tire, the amount of corrective force required is more smoothly related to the angle of the lean, and you might be able to make small lean adjustments better.
ALSO! I'm not certain, but I think the shape of the tire miiight have an effect on how the lean angle translates into turning as the wheel rolls along. I think a rounder tire might be more prone to turn towards the lean, which would be helpful, but I'm not sure.
Some considerations from personal experience in prototyping similar stuff: 1. use large hex core patterns for the gear sectors. Generally speaking, if you can get a 8 mm hex to a 2 cm wide part, you're set with a great ratio on a normal wall thickness. I've wall hangers that are built that way and they can soak up to 15kgs despite being very thin in section (no machine figuring involved, all hand laid lattice). 2. Instead of making recesses with small bits to joint the sections, which will weaken the whole structure, make another exterior ring that has overlapping segments (like brickwork). The equal sections will make for great structure. If you print this outer ring in two sections, so a middle wall (which was the base of the print) is shared, the structure jumps in structure even more. -that outer ring, give it a ridge that's dovetailed to capture the TPU tire, which you can then print in a spiral, one piece, and slide over the dovetail. The tire would be in two pieces, so you can print the dovetail halves properly, then glued using Loctite's black rubber glue (pliable, but extremely strong, strong enough to glue a tire to a rim, on a car, permanently). You will not need any kind of screw for this, it can be build as a slot setup, which is far better than screws, which require you make a structure in structure when printing, to give you a strong boss. 3. Finally, if a gear sticks, it's because the tolerances aren't right from the print, and rather than printing another piece, just print yourself a master tooth, that's one sandpaper thickness smaller than the actual tooth valley you're trying to sand away. Then a little bit of elbow grease will yield perfect gear teeth. If you have an oscillating type dremel, the kind usually used to sand seam lines in model making, you can screw this tooth + sand paper to the bit, and sand that way. -adding superglue with a brush before hand, will give you extremely strong and smooth teeth. Less rolling resistance, less noise (which itself is a loss), and generally speaking, smoother operation.
@@benkaye5014 Sorry for late reply. Instead of letting the slicer fill your piece (depending on how much material you want to print) you take hexagons and place them in a grid leaving a bit more meat, like say for a part that's the size of a cup, you put 1 mm thickness walls. You make this lattice much larger than the part you want, then you place it so that the hexagon tubes formed by it are placed parallel to the direction you want the least deformation. Here's the crucial part. Every hexagon that's not full, you have to fill out, which a slicer will not. Learnt this from a guy in a prop making forum. Makes extremely strong components.
Even with my fairly large lack of understanding of the build it still absolutely genius and the video was as awesome as ever, fairly new~ here and I’ve loved everything I’ve seen you create It is also all highlighted by how passionate you are for your craft, much respect!
You could either make sure the whole inside of the robot stays upright (and use a separate weight for leaning), or put the control-moment-gyros on a gimbal (or even two gimbals) so they can stay upright while the rest of the robot leans. Also maybe make a case for the gyro so you don't accidentally touch it and lose skin when it's spinning fast :P
That's a very clever use of bearing balls to get your mass. Perhaps you can use an RPM differential on your speed controllers to get side loading on your gyro to help offset the twist when you try to turn.
I caught your video as I was surfing UA-cam. I like how your mind works. There was a moment at the end however, where you seam to be perplexed as to why it turns the opposite way of the lean. I think this has to do with how bicycles turn. While riding along, if you want to turn Left, you pull back slightly on the Right end of the handle bar. This moves the lower part of the bike off center to the Right. As the bike leans into the turn the handle bars, and the front tire, turn to the Left to follow the circle you're riding in. That's just my guess, but it seams to make sense to me.
For the motors that are used to stabalize the main wheel side to side, you could use another gimbal set on the horizontal axis to stabilize the gimbals in the vertical axis.
Gyroscopic forces are fun! In motorcycle racing you are taught to push on the right handlebar if you want to lean and turn right. I thought the instructor was nuts, but it works! When your project fell over, all I could think was... "Nooo! Not the nice shiny floor!" lol Wonder if you could create an inner carriage, to mount the gyro's, that is independent of everything else. Then the drive carriage could have the drive motor, gearing, controls, etc. A lot of parts, but would help with the gyro axis changing problem.
I'm really curious how much a yaw reaction wheel would improve performance, especially at higher speeds. Then you could more directly twist the wheel like a unicyclist does. I think as the ground speed increases, you could use the reaction wheel to do more and more of the stabilization, especially if you're not too worried about which direction it ends up going. You might even be able to use it to offset some of the off-axis gyroscopic forces when they're tilted.
Wow, this is a really tricky control problem. These videos are making me ache for an elegant solution that keeps the wheel stable and allows for controlled turns
Wow. I thought a lot about this idea. Great to see you actually build it. I would have tried to make the gyros nearly the same size as the wheel so you can still sit inside.
It feels like for turning you could use a servo on top with some mass and turn it in the opposite direction you would like the vehicle to turn. Great project as always
Hi James. I noticed the wheel surface is quite flat. When you are driving a motorcycle and your tires wear out they also get a flat surface. Getting to turn the motorcycle then I way more.difficult. you especially notice it when you put a brand new tire.on. it's almost like it's effortless. I'm not sure trying to balance the wheel when it's not moving is the right way of thinking. It's about your active balancing mechanisms to balance it. And when there is a big flat it makes the response not lineair
Steering is a true conundrum! Motorcycle racers know the secret. There's a great vid by the Veritasium channel called 'Most People Dont Know How Bikes Work' that explains this phenomenon. I hope you build this into your monowheel.
Another amazing build.. Maybe a good alternative to gyroscopes here might be a simple moving balance weight. Maybe the weight could be made as a vertically mounted momentum wheel, a fairly painless way to achieve steering. (just realized that you still end up with the same problem but now rotated through 90 degrees.)
I think weight's with stabilisers is good alternate indeed. The one-wheel motorcycle from "yugioh jack atlas" works simulair with LOW place mass around the main wheel
I recommend to check out electric unicycles. They are monowheels of a sort too, but the rider stands on them instead. Very useful and fun, bigger ones have enough range to go over 100km on a charge with speeds in excess of 60km/h
@@Blox117 "of a sort" like i said. Though electric unicycles are also often referred as monowheels and they are actually useful for everyday life. I use mine to commute daily. I get to work much faster than with a car or a train.
@@MikaelLevoniemi monowheels specifically refer to a single track wheel comprising the entirety of the vehicle with the components in the middle. a uni rides on a wheel like any other vehicle. a monowheel has no axle or hub like a regular wheel does
@@Blox117 Yes? And that's exactly how electric unicycles are built. Many of the components are inside the wheel, except the batteries are on the sides. Many models don't even have an axel because the motor handles all that. There are no mechanical brakes etc. There's a good reason why electric unicycles are often called monowheels. The difference is in size and rider location. Example of an EUC in use in Paris ua-cam.com/video/wTKowa7tqsk/v-deo.html I use mine to commute and it's much faster than a car or walk+train.
You need a horizontal gyro, it would add mass and also allow to rotate the whole mechanism and have the added bonus of providing a counter force to the whole wheel rotation.
I guess there are two other solutions: Have a heavy ring running inside the wheel arch as a flywheel. So if you decelerate, you just accelerate the flywheel's mass the other way around, bleeding off the energy. This would allow you even to reuse the energy, when you want to accelerate again. And you can also accelerate fast and spin the flywheel in the other direction to avoid tilting. This energy could either be recovered while driving or you can just use it up when you want to brake. The other solution would be to have a flywheel in the center but the axle turned 90° horizontally - so it will face sideways. You would need to spin up this flywheel to a very high speed, but given that it is heavy enough it will stabilise the tilting action pretty well. You could even use it to stabilise the tilting right / left by speeding it up/slowing it down. So you can steer and stabilise it. The only weird thing would be, that if the center portion do lean forwards or backwards it will stay there. So maybe you could combine both solutions
Interesting. Optimal sideways balance seem to require that the gyro mechanism is not leaning in forward-backward direction and should be located somewhat centrally to get that sideways torque. On the other hand forward-backward motion is based totally on gravity only - the lower the center of mass of a frame that turns the wheel the better the control would be. Seems difficult to achieve these with singular inner part to the wheel but what if you decouple these things into two inner parts? The gyro part keeps itself upright no matter if the wheel is turning or not and another weighted part takes care of forward-backward motion.
I'd add a line or two of code to taper power to the motors if it falls over. It looks like it might sever a hand when you try to stop it spinning out of control. Would giving the gyros an internal frame to move in, to keep them 90% horizontal, limit the effects they cause in other planes when in forward motion? That was the only theoretical solution I could think of but adding that adds another motor/battery for the horizontal gyro stabilisation adds mass. Looking forward to the solution to this tho.
Great work. A couple of quick suggestion for future builds: Program one of the toggles on your controller as a soft emergency stop that kills the motors Program a ramp speed for the gyro speed controllers so you don't have to manually ramp them up.
Look up counter steering on motorcycles. Simply put, the term 'countersteering' refers to the principle (governed by the laws of physics!) that to turn a moving motorcycle in a given direction, the rider must turn the handlebar in the opposite direction of the turn. Sounds exactly like the phenomenon you're experiencing turning your mono wheel.
Look up how you steer a motorbike and that will help with the concept of the wheel leaning left but going right. :) it’s called counter steering I believe.
Maybe try a design with the gyros in the center of the circle. That way even if you go forward the gyros won't move from the center of mass of the geometrics of the balanced wheel. Within this desing the balance of the gyros are high up the center, in that case when you move forward or backwards you will always create an inclination where after a certain angle the gyros wouldn't compensate and then the precession of the angular momentum is going to overcome the balancing action. Or maybe try to add another degree of freedom for the gyros, where they can balance in perpendicular to the movement of the wheel. That's a nice project, incredible video.
You're explanation for going round the corner is in fact an old problem that every human solved by feeling. You do it differently than you would expect! If you want to go to the right you will steer left at first! By doing so, you "fall" in the right direction and catching yourself from falling over. You're code should implement this and everything will work fine. There is an old experiment where they try to run a motorcycle by a robot, what ever they could think of it could only run in a straight line. Finally they measured a human driving the same motorcycle and found the slight movement in the wrong direction. It is also known as counter steering.
I believe that in order to steer to the right you should first steer slightly to the left, so that the monowheel "falls" a little bit to the right and then you can steer to the right without falling to the left and continue your steering movement to the right. . Hope it makes sense
I notice when it falls over, you struggle to quickly pick it up before too much damage can occur to the floor and gyro wheel. Perhaps having a static cover plate over them that is attached at the top and bottom pivot brackets might help prevent that, as well as prevent it from dancing around as the gyro wheel contacts the ground.
Add a third gyro that spins in the horizontal plane. I'm not sure where you could squeeze that in, though. Maybe adding a second axis of tilt to the two existing gyros would also work?
For your gyro steering to steer it in your desired direction you first need to initiate a turn in the wrong direction, this will cause it to lean in your desired direction, and then turn into the lean in order to stabilize it, turn even more into the lean to pull out of the turn. It is comparable, although not totally similar, to how you need to initiate a turn for a bike with 2 wheels where only the front wheel turns.
@@Blox117 1. No it isn't, while easier to observe in a two wheeled system it is the offset of the balance point relative to the steer point that causes the effect. Moving the steer point in front of the balance point should negate the effect somewhat. 2. The speed at which this comes into affect varies in relation to the size of the system. It becomes relevant when moving at fairly slow speeds, and stays relevant even at very high speeds. it starts to become relevant around 0.001 kmh at the size of a marble, and around 0.1 kmh at the size of a bicycle.
@@OmegaWoodworm im guessing you've never stepped foot out of a 4 wheeled steel box, have you? enjoy knowing that i am able to magically turn without doing any of what you stated lmao.
@@Blox117 I see you have missed some context. The effect I am referring to affects 1 wheeled and 2 wheeled systems, not 4 wheeled ones. Since cars have more than 2 wheels they are not subject to this effect.
Put the axle of the gyros in the centre of the wheel. That way they won't be moved forward or backward when the internals are leaned forward or backward to move. Use static weights, like those unconnected batteries, to get it to move when leaning forward or backward.
I've always wanted to build an RC monowheel but with just a linear actuator moving a mass side to side to steer and keep balance. If this mass was positioned at the wheel's axis of rotation, I believe it would not cause any unwanted precession. As the whole wheel is effectively a gyro, one should be able to get it stable when running.
I notice that the gyroscope axis of rotation is above the wheel centerline. Would aligning the two help your stability problems? I would think this would help with the change of force from the gyros when tilting the robot forward and backward.
James Bruton DUDE!!!!!! Do you know about Electric UniCycles (EUC)? You need to. Currently the EUC market is flooded with monowheels that only have 180 degrees of self balancing: the wheel self balances from front to back (like a Segway). What’s missing is side to side. Current riders have to control and maintain side to side balance. But if the EUC world had a full 360 degree Gyro Wheel, that would be a huge game changer. Reach out to companies like Inmotion or Begode with your ideas and knowledge. We need this in the Electric UniCycle community…NOW. Great work. Keep it up!
I love seeing videos where things don't work as expected. Keeps us in reality and helps us understand the number of failures that it takes to get a success! Awesome project. I don't seem to see it anywhere, but what are the joysticks that you are using on your remote?
Great Build!!! (as always) only one small note, you know this effect if you have ever ridden a motorcycle or bicycle, I believe it is the same reason here as well that you need to lean the other way than the direction you are turning, which is also called counter steering, even though it has 1 wheel, as you implemented the gyroscopic wheels spin by the brushless motors, the overall system behaves like a bike or motorcycle instead of a monowheel. It seems to be a little janky right now but it seems to be fixable as well, I am really hyped about this, could you make one with a more common (easier to find and cheaper) parts version of this, the scale can change but again with gyro wheels? That would be great!
will be nice to see it built with the gyro having an additional axis. They run the whole thing at some speed, as you did in the carpark with the previous version. Should be quite impressive.
A small little teny tiny suggestion James. 11:18 please please start adding panic button to your creations. You know the one button you push and it disables everything.
When you tried the leaning steering, it behave like a bike. A bike must be first be countersterd to initiate a leaning and then steered as required. Maybe there is a similar thing for this?
To solve the issue you described at the end of the video, would it be possible to have a mass at the top of the wheel move forwards and backwards to act as the forwards/backwards controls? This would allow the body to stay upright and avoid the gyro steering issue I think.
The offset center where it counter steers reminds me very much of how a bicycle front wheel steers, whenever a cyclist turns right, it first swings the steering bar slightly left before turning right.
you need two more gyros on the top and bottom within the wheel itself to counteract movement in the sideways direction that it falls over in. if you have 4 gyros you will be able to drive it.
As James mentioned, part of the problem is that the gyroscopes are leaning forward. I'd bet it would be less of a problem if it leaned backward. This is how bicycles' front wheels stay straight when you take your hands off the handle bars. :-)
Wouldn't it be better if you place the accelerometer closer to the center of the robot? I've seen many projects where you place it in kinda random places. That may impact the correctness of measurements and require compensations in the PID and control loop
Hey James i have fell of you're videos fore a couple of years. you remember you're first big project. BB-8 v1. This is wat wanted to do when you struggled with balance. 😁. Lowe you're awesome engineering i want to learn to do stuf like you.
instead of changing the pressure angle on the spur gear, try increasing the backlash that way you have a wider area of contact to reduce wear on the gear while still having some extra space for flex in the gear
Seems like you should be able to use the turning angle to alter the gyroscope's center point so that when you "steer" it knows that the center of your angle is meant to be offset from your true vertical center. However, if you use the boat analogy, you should also be able to tilt your gyroscope physically if you place it on a horizontal rig.
could you could you make the gyro section tilt the opposite direction of the forward backward lean so that it would always be upright?
2 роки тому
I'm wondering if the profile of the wheels that are causing issues with steering. When the robot leans slightly left the contact point moves quite far to the left, causing the actual balance to shift towards the right. Maybe try with a rounder profile for the wheels.
For the record, I know there are many liability concerns with a project like this. My newest wheelchair is from a small startup company. They were able to get around the regulations by simply marketing the device as a personal EV. If the item is not marketed towards the disabled, it is completely unregulated. This is how most of the smaller companies get their product to market
Wow - this is very complex to try it with the side-to-side balancing Gyro deflection axis' constrained to match the forward-back balance axis which, also is even further constrained to the forward-back drive force angle by virtue of balance-bots only having a single forward-back movement "joint". I'm not smart enough to visualize how those PIDs would work together as a whole but, intuitively, it doesn't seem possible without decoupling at least one of those tied axis... It would be mechanically more complex to nest that Gyro unit inside an internally balancing ring to isolate the Gyro's vertical axis' but doing so would be dramatically simpler from a systems integration point-of-view since you could then initially work on tuning the bank and pitch PIDs separately. If you also add an internally isolated Yaw Gyro pair, then you could steer it without roll/bank coordinated turns. Not likely your end goal. but once it can keep functioning and has a means of brute-force stumble recovery then there is the option of handing it off to a ML project to teach itself to coordinate the axis' and roll efficiently. Super Fun Project and thought exercise! Thanks for doing all the hard work!
Putting the gyros in the wheel's center and putting them on their own race should fix a handful of issues. Centering them would disallow the tilting and turning induced by momentum, and putting them on their own race would keep the upright axis of the servo control. You also need some form of momentum dumping to desaturate the gyros anyways. If you really want a top-weighted wheel, to poorly emulate leverage force that a unicycle would see, I would suggest flipping your current configuration upside down, have the drive motor and batteries at the top, though most of this leverage will be made irrelevant by the heavy gyros in the center of the wheel; tilt to drive just isn't appropriate unless you have a significant lever on the robot.
Would it help to let the drive motor work through a differential and use a servo on the body of the diff for fine tuning while you're standing still? The motor is not agile enough for small compensation and causes overrunning the zero point. It does then introduce yet another mechanism... Just my 2 cents
Hey James. Why are your gyroscopes not always centered on the robot? You can build gyroscopes with bearings. Then it will always be in a horizontal position and can stabilize the position.
What if the gyros had their own bering assembly so the robot can rotate around them when driving forward and the gyros can stay perpendicular to the ground.
Very inspiring. I still cannot understand, how you can make so many videos with such high quality robot builds. Impressive!
Yo nice seeing you here man! Love your vids
Heyyyy 🤘
with a fast 3d printer and even faster brain 😍
hello, you are both great scientists!kisses
He must be putting in about 12 hour days. It takes me all day just to figure out the bugs in my code haha
this is why spacecraft that use control moment gyros for stabilization always have some other way of holding attitude. if the spacecraft is subject to some unbalanced torque (from radiation pressure, magnetic fields, etc) the CMGs will move as far as they can go and become saturated. to unsaturate them, they're moved back to the neutral position and the spacecraft's thrusters are used to counteract the resulting torque. the term for this is momentum dumping. the robot crashes because it has no way of doing this. being tilted over results in a constant torque which will quickly begin to saturate the gyros, so that they will be moved very far in one direction even when the robot returns to vertical. unless you plan on doing the exact same amount of left turns as right, this concept fundamentally cannot work. now, this *does* give me the idea of creating some kind of robot that uses CMGs for most of its control, but has compressed air jets for momentum dumping.
The robot could unsaturate the reaction wheels by leaning a small amount the opposite way and letting gravity unsaturate the wheels (I think, not entirely sure). But yeah, it has to be done standing still if you don't want to turn.
So, a robot with corrective farts? Sign me up!
@@diegodoumecq5144 ... #JustAddThruster
@@huvudpersson7344 pretty sure this will work. You just need a bit gyro power left to flip the leaning side and let it hover there until desaturated. would be fun to implement this in the stabilize algorithm and correct its saturation levels as it goes.
To solve the forward backward leaning problem of the gyro add them to a slide that corrects its position back to center. Could even be used to help control steering. I would love to work on the algorithms doing all that in parallel. Must be fun to watch afterwards doing its things to balance.
How do boats do momemtum dumping?
Wheels within wheels ...
Perfect example of how adding two mechanisms you have experience with can lead to completely new and interesting problems
There's only one thing left to do....a no wheel balancing robot. Great skills though and best wishes >1M subscribers
And then a no robot balancing robot, so two things.
@@4LXK 😄 I reckon I could build one of them
Hold on, first try 0.5 wheels before we dive in xD
I would love a pogo-stick type platform that could carry a person. It can handle stairs and irregular terrain, so an alternative to a wheel chair or stroller.
@@JohnDlugosz Build a huge robo dog
Most people don't know how bicycles actually work, to turn left, you first have to turn right, to change your cog to allow you to then turn left, Veritasium did a great video on this. Seems like your monowheel is experiencing this exact same mechanic.
nope you dont have to turn at all. dont repeat information if you never actually did it yourself. countersteering is useful for making sharp turns but is not how bicycles always turn
If you're riding a motorcycle at speed, you must point the wheel to the left to turn right. Pointing left causes the mass to lean right, this turning right. The wheels are the gyros.
Bicycles and motorcycle actually respond the same. If you're moving parking lot speed, left is left, and right is right. But as your speed increases, the more influence from the gyro, inverting your correlation. If you get your bicycle up to 30mph+, you will experience this.
I'm very impressed, even if the result itself isn't performing too well. There is so much to learn from your designs.
You need a slower PID that takes the gyro position as input and changes the left right angle setpoint as it's output. That way it will lean itself left and right automatically to keep the gyros from hitting their end stops.
Also the dampening of the pid is to weak for the side to side stabilization.. it oscillates quit heavily 😅
Actually PID is not a good control strategy. The force vector being so close to the center of mass causes a bifurcation in the systems phase portrait. The system is not guaranteed to have unique control solutions.
So the PID assumes solution uniqueness, i doubt PID is sufficient. lowering the center of mass more will make the pid work better
13:03 did you see Veritasium’s recent video about steering? I wonder if you need to apply the same principle. In his video he shows how a bicycle has to steer the WRONG way first (just a little bit and very quickly) before steering the RIGHT way. It’s fascinating and might be helpful!
I was just going to say the same thing.
Nice connection, although it doesn’t really apply here since you reverse steer to get the bike tilted in the direction you want to steer, and here you use the gyro to control the tilt. Imagine if you added a gyro to a bike, then you would never need to reverse steer since the gyro takes the role away from the front wheel. Actually you wouldn’t need to steer the front wheel at all but I imagine that would be a very difficult bike to ride.
You are right about the gyros, assuming they interface with the bodies containing them with torque around every axis.
This is however not the case since the gyros here spin in the "same" plane as the wheels. They regulate pitch in that manor only.
It is only when you offset the planes the gyros rotate in that the resulting forces are exerted push side to side.
In the bike case the force is applied below the center of mass causing the bike to tilt towards the direction you want to turn. (The bike rides out from under you initially)
If this force is applied below the center of mass it is likely similar to the bike but less severe.
Take note that the two monowheels regulate pitch differently because of their location with respect to the center of mass. This supports the idea that it is indeed what vertasium describes.
@@molomono9481
>This is however not the case since the gyros here spin the the “same” plane as the wheels.
The way the gyro is used doesn’t matter in my explanation, as a reaction wheel or control moment, as long as it controls the tilt left and right.
>If this force is applied below the center of mass it is likely similar to the bike but less severe.
For gyros their position relative to the center of mass doesn’t matter, just the orientation, which is counter intuitive. Although you are right that the unwanted steering is somewhat related to the center of mass.
It tries to steer the wrong way because the wheel is pitching down while turning (center of mass at top), which tilts the roll axis downward so when the gyro tries rolling right (to offset gravity when turning left) it spins the wheel to the right at the same time.
If the wheel was pitching up to drive forward (center of mass on bottom), however, then the reverse would happen and the axis tilt would actually help with the turn.
So the center of mass affects which way the wheel pitches to go forward, which affects how the roll axis is tilted, which affects turning.
@@corporalwill123 haha yeah we agree on the steering but are focussing on different interfaces of force in the system to explain it. Would be interesting to play with both variables and study what is really going on
Its weird though since you would expect that the gyro exerting force normal to the center of mass in xy plane would be simple since it is removing non-linearity but it is actually near a bifurcating region making linearized control have non unique control solutions.
In other words, PID wont work well
constantly impressed by your contraptions, Mr. Bruton.
Random project idea: make Gizmoduck from the show “Ducktales.” I loved that cartoon as a kid and ever since I discovered your channel and the possibilities of 3D printed self-balancing robots I’ve want to see a real robot duck riding around on a single wheel. As always thanks for your great content.
I think tire shape might be a sneakily important factor. A flat-bottomed tire might seem easier to stabilize when the wheel is stationary and near vertical, but once you tip over the edge, the gyros are bound to struggle at correcting the sudden change in forces. Plus, when the wheel is moving, it's gonna be hard to make minor steering adjustments (like the video mentioned) since the gyros' smaller adjustments don't really lean the wheel at all. With a rounder tire, the amount of corrective force required is more smoothly related to the angle of the lean, and you might be able to make small lean adjustments better.
ALSO! I'm not certain, but I think the shape of the tire miiight have an effect on how the lean angle translates into turning as the wheel rolls along. I think a rounder tire might be more prone to turn towards the lean, which would be helpful, but I'm not sure.
Beautiful, my brain was reeling before you even got thru the intro !
You are Full Stark !
This is why I like science, engineering, electronics, and diy rc builds.
As someone obsessed with the idea of sports-bike-like monowheels, these two videos gave me so much insight into it you can't imagine...
Some considerations from personal experience in prototyping similar stuff:
1. use large hex core patterns for the gear sectors. Generally speaking, if you can get a 8 mm hex to a 2 cm wide part, you're set with a great ratio on a normal wall thickness. I've wall hangers that are built that way and they can soak up to 15kgs despite being very thin in section (no machine figuring involved, all hand laid lattice).
2. Instead of making recesses with small bits to joint the sections, which will weaken the whole structure, make another exterior ring that has overlapping segments (like brickwork). The equal sections will make for great structure. If you print this outer ring in two sections, so a middle wall (which was the base of the print) is shared, the structure jumps in structure even more.
-that outer ring, give it a ridge that's dovetailed to capture the TPU tire, which you can then print in a spiral, one piece, and slide over the dovetail. The tire would be in two pieces, so you can print the dovetail halves properly, then glued using Loctite's black rubber glue (pliable, but extremely strong, strong enough to glue a tire to a rim, on a car, permanently). You will not need any kind of screw for this, it can be build as a slot setup, which is far better than screws, which require you make a structure in structure when printing, to give you a strong boss.
3. Finally, if a gear sticks, it's because the tolerances aren't right from the print, and rather than printing another piece, just print yourself a master tooth, that's one sandpaper thickness smaller than the actual tooth valley you're trying to sand away. Then a little bit of elbow grease will yield perfect gear teeth. If you have an oscillating type dremel, the kind usually used to sand seam lines in model making, you can screw this tooth + sand paper to the bit, and sand that way.
-adding superglue with a brush before hand, will give you extremely strong and smooth teeth. Less rolling resistance, less noise (which itself is a loss), and generally speaking, smoother operation.
can you explain the hex core pattern for gear sector?
@@benkaye5014 Sorry for late reply. Instead of letting the slicer fill your piece (depending on how much material you want to print) you take hexagons and place them in a grid leaving a bit more meat, like say for a part that's the size of a cup, you put 1 mm thickness walls. You make this lattice much larger than the part you want, then you place it so that the hexagon tubes formed by it are placed parallel to the direction you want the least deformation.
Here's the crucial part. Every hexagon that's not full, you have to fill out, which a slicer will not.
Learnt this from a guy in a prop making forum. Makes extremely strong components.
Even with my fairly large lack of understanding of the build it still absolutely genius and the video was as awesome as ever, fairly new~ here and I’ve loved everything I’ve seen you create
It is also all highlighted by how passionate you are for your craft, much respect!
You could either make sure the whole inside of the robot stays upright (and use a separate weight for leaning), or put the control-moment-gyros on a gimbal (or even two gimbals) so they can stay upright while the rest of the robot leans. Also maybe make a case for the gyro so you don't accidentally touch it and lose skin when it's spinning fast :P
Can I just say, I love the music you use for showing the 3D printing!
That's a very clever use of bearing balls to get your mass. Perhaps you can use an RPM differential on your speed controllers to get side loading on your gyro to help offset the twist when you try to turn.
It amazes me how easy you make all this look, knowing it's anything but.
I caught your video as I was surfing UA-cam. I like how your mind works. There was a moment at the end however, where you seam to be perplexed as to why it turns the opposite way of the lean. I think this has to do with how bicycles turn. While riding along, if you want to turn Left, you pull back slightly on the Right end of the handle bar. This moves the lower part of the bike off center to the Right. As the bike leans into the turn the handle bars, and the front tire, turn to the Left to follow the circle you're riding in. That's just my guess, but it seams to make sense to me.
For the motors that are used to stabalize the main wheel side to side, you could use another gimbal set on the horizontal axis to stabilize the gimbals in the vertical axis.
Gyroscopic forces are fun! In motorcycle racing you are taught to push on the right handlebar if you want to lean and turn right. I thought the instructor was nuts, but it works! When your project fell over, all I could think was... "Nooo! Not the nice shiny floor!" lol Wonder if you could create an inner carriage, to mount the gyro's, that is independent of everything else. Then the drive carriage could have the drive motor, gearing, controls, etc. A lot of parts, but would help with the gyro axis changing problem.
I'm really curious how much a yaw reaction wheel would improve performance, especially at higher speeds. Then you could more directly twist the wheel like a unicyclist does. I think as the ground speed increases, you could use the reaction wheel to do more and more of the stabilization, especially if you're not too worried about which direction it ends up going. You might even be able to use it to offset some of the off-axis gyroscopic forces when they're tilted.
Wow, this is a really tricky control problem. These videos are making me ache for an elegant solution that keeps the wheel stable and allows for controlled turns
Wow. I thought a lot about this idea. Great to see you actually build it. I would have tried to make the gyros nearly the same size as the wheel so you can still sit inside.
Congratulations 🎉🎈🎊🍾 1 million subs
I was waiting for this …you deserve
مبارك المليون
That is so elegant. Nice one James...
It feels like for turning you could use a servo on top with some mass and turn it in the opposite direction you would like the vehicle to turn. Great project as always
Hi James. I noticed the wheel surface is quite flat. When you are driving a motorcycle and your tires wear out they also get a flat surface. Getting to turn the motorcycle then I way more.difficult. you especially notice it when you put a brand new tire.on. it's almost like it's effortless. I'm not sure trying to balance the wheel when it's not moving is the right way of thinking. It's about your active balancing mechanisms to balance it. And when there is a big flat it makes the response not lineair
Steering is a true conundrum!
Motorcycle racers know the secret.
There's a great vid by the Veritasium channel called 'Most People Dont Know How Bikes Work' that explains this phenomenon.
I hope you build this into your monowheel.
Another amazing build.. Maybe a good alternative to gyroscopes here might be a simple moving balance weight.
Maybe the weight could be made as a vertically mounted momentum wheel, a fairly painless way to achieve steering.
(just realized that you still end up with the same problem but now rotated through 90 degrees.)
I think weight's with stabilisers is good alternate indeed. The one-wheel motorcycle from "yugioh jack atlas" works simulair with LOW place mass around the main wheel
Awesome project! Why did you use the Nano to monitor the IMU rather than just using the Mega for both the servo and the IMU?
I recommend to check out electric unicycles. They are monowheels of a sort too, but the rider stands on them instead. Very useful and fun, bigger ones have enough range to go over 100km on a charge with speeds in excess of 60km/h
its a uni not a monowheel
@@Blox117 "of a sort" like i said. Though electric unicycles are also often referred as monowheels and they are actually useful for everyday life. I use mine to commute daily. I get to work much faster than with a car or a train.
@@MikaelLevoniemi monowheels specifically refer to a single track wheel comprising the entirety of the vehicle with the components in the middle. a uni rides on a wheel like any other vehicle.
a monowheel has no axle or hub like a regular wheel does
@@Blox117 Yes? And that's exactly how electric unicycles are built. Many of the components are inside the wheel, except the batteries are on the sides. Many models don't even have an axel because the motor handles all that. There are no mechanical brakes etc. There's a good reason why electric unicycles are often called monowheels. The difference is in size and rider location. Example of an EUC in use in Paris ua-cam.com/video/wTKowa7tqsk/v-deo.html
I use mine to commute and it's much faster than a car or walk+train.
You need a horizontal gyro, it would add mass and also allow to rotate the whole mechanism and have the added bonus of providing a counter force to the whole wheel rotation.
I guess there are two other solutions:
Have a heavy ring running inside the wheel arch as a flywheel. So if you decelerate, you just accelerate the flywheel's mass the other way around, bleeding off the energy. This would allow you even to reuse the energy, when you want to accelerate again.
And you can also accelerate fast and spin the flywheel in the other direction to avoid tilting. This energy could either be recovered while driving or you can just use it up when you want to brake.
The other solution would be to have a flywheel in the center but the axle turned 90° horizontally - so it will face sideways.
You would need to spin up this flywheel to a very high speed, but given that it is heavy enough it will stabilise the tilting action pretty well.
You could even use it to stabilise the tilting right / left by speeding it up/slowing it down. So you can steer and stabilise it.
The only weird thing would be, that if the center portion do lean forwards or backwards it will stay there.
So maybe you could combine both solutions
Interesting. Optimal sideways balance seem to require that the gyro mechanism is not leaning in forward-backward direction and should be located somewhat centrally to get that sideways torque. On the other hand forward-backward motion is based totally on gravity only - the lower the center of mass of a frame that turns the wheel the better the control would be. Seems difficult to achieve these with singular inner part to the wheel but what if you decouple these things into two inner parts? The gyro part keeps itself upright no matter if the wheel is turning or not and another weighted part takes care of forward-backward motion.
I'd add a line or two of code to taper power to the motors if it falls over. It looks like it might sever a hand when you try to stop it spinning out of control.
Would giving the gyros an internal frame to move in, to keep them 90% horizontal, limit the effects they cause in other planes when in forward motion? That was the only theoretical solution I could think of but adding that adds another motor/battery for the horizontal gyro stabilisation adds mass. Looking forward to the solution to this tho.
Great work. A couple of quick suggestion for future builds:
Program one of the toggles on your controller as a soft emergency stop that kills the motors
Program a ramp speed for the gyro speed controllers so you don't have to manually ramp them up.
Look up counter steering on motorcycles. Simply put, the term 'countersteering' refers to the principle (governed by the laws of physics!) that to turn a moving motorcycle in a given direction, the rider must turn the handlebar in the opposite direction of the turn. Sounds exactly like the phenomenon you're experiencing turning your mono wheel.
Look up how you steer a motorbike and that will help with the concept of the wheel leaning left but going right. :) it’s called counter steering I believe.
Maybe try a design with the gyros in the center of the circle. That way even if you go forward the gyros won't move from the center of mass of the geometrics of the balanced wheel. Within this desing the balance of the gyros are high up the center, in that case when you move forward or backwards you will always create an inclination where after a certain angle the gyros wouldn't compensate and then the precession of the angular momentum is going to overcome the balancing action.
Or maybe try to add another degree of freedom for the gyros, where they can balance in perpendicular to the movement of the wheel.
That's a nice project, incredible video.
You're explanation for going round the corner is in fact an old problem that every human solved by feeling. You do it differently than you would expect! If you want to go to the right you will steer left at first! By doing so, you "fall" in the right direction and catching yourself from falling over. You're code should implement this and everything will work fine. There is an old experiment where they try to run a motorcycle by a robot, what ever they could think of it could only run in a straight line. Finally they measured a human driving the same motorcycle and found the slight movement in the wrong direction. It is also known as counter steering.
Whenever the 3d printing music comes on, I imagine I'm watching the weather channel.
I believe that in order to steer to the right you should first steer slightly to the left, so that the monowheel "falls" a little bit to the right and then you can steer to the right without falling to the left and continue your steering movement to the right. . Hope it makes sense
Awesome! Its so close to being the coolest monowheel rc out there !
This really is very impressive James!
How many sponsors do you want?
Bro was like: Yes
I notice when it falls over, you struggle to quickly pick it up before too much damage can occur to the floor and gyro wheel. Perhaps having a static cover plate over them that is attached at the top and bottom pivot brackets might help prevent that, as well as prevent it from dancing around as the gyro wheel contacts the ground.
Add a third gyro that spins in the horizontal plane. I'm not sure where you could squeeze that in, though. Maybe adding a second axis of tilt to the two existing gyros would also work?
For your gyro steering to steer it in your desired direction you first need to initiate a turn in the wrong direction, this will cause it to lean in your desired direction, and then turn into the lean in order to stabilize it, turn even more into the lean to pull out of the turn.
It is comparable, although not totally similar, to how you need to initiate a turn for a bike with 2 wheels where only the front wheel turns.
1. this information is incorrect
2. the speed at which this comes into play is much higher than achieved here
@@Blox117
1. No it isn't, while easier to observe in a two wheeled system it is the offset of the balance point relative to the steer point that causes the effect. Moving the steer point in front of the balance point should negate the effect somewhat.
2. The speed at which this comes into affect varies in relation to the size of the system. It becomes relevant when moving at fairly slow speeds, and stays relevant even at very high speeds.
it starts to become relevant around 0.001 kmh at the size of a marble, and around 0.1 kmh at the size of a bicycle.
@@OmegaWoodworm 1. yes it is, you are just ignorant
@@OmegaWoodworm im guessing you've never stepped foot out of a 4 wheeled steel box, have you? enjoy knowing that i am able to magically turn without doing any of what you stated lmao.
@@Blox117 I see you have missed some context.
The effect I am referring to affects 1 wheeled and 2 wheeled systems, not 4 wheeled ones.
Since cars have more than 2 wheels they are not subject to this effect.
Nice work - a good example for engineering and find a solution! Like it.
Put the axle of the gyros in the centre of the wheel. That way they won't be moved forward or backward when the internals are leaned forward or backward to move. Use static weights, like those unconnected batteries, to get it to move when leaning forward or backward.
words cannot describe how much i have learned in 15 minutes.
and how ill abuse my new knowledge to make others heads hurt.
It's always great to see naomi wu in one of your videos :)
I've always wanted to build an RC monowheel but with just a linear actuator moving a mass side to side to steer and keep balance. If this mass was positioned at the wheel's axis of rotation, I believe it would not cause any unwanted precession. As the whole wheel is effectively a gyro, one should be able to get it stable when running.
Absolutely amazing! Keep up the remarkable work.
I notice that the gyroscope axis of rotation is above the wheel centerline. Would aligning the two help your stability problems? I would think this would help with the change of force from the gyros when tilting the robot forward and backward.
9:00 isn't it easier to just design a circuit to control inrush current rather than having to fiddle with the speed manually?
James Bruton DUDE!!!!!! Do you know about Electric UniCycles (EUC)? You need to. Currently the EUC market is flooded with monowheels that only have 180 degrees of self balancing: the wheel self balances from front to back (like a Segway). What’s missing is side to side. Current riders have to control and maintain side to side balance. But if the EUC world had a full 360 degree Gyro Wheel, that would be a huge game changer. Reach out to companies like Inmotion or Begode with your ideas and knowledge. We need this in the Electric UniCycle community…NOW. Great work. Keep it up!
It's this: ua-cam.com/video/lS_yyy0gmUQ/v-deo.html
someone that actually deserves a subscribe
I love seeing videos where things don't work as expected. Keeps us in reality and helps us understand the number of failures that it takes to get a success! Awesome project.
I don't seem to see it anywhere, but what are the joysticks that you are using on your remote?
Great Build!!! (as always) only one small note, you know this effect if you have ever ridden a motorcycle or bicycle, I believe it is the same reason here as well that you need to lean the other way than the direction you are turning, which is also called counter steering, even though it has 1 wheel, as you implemented the gyroscopic wheels spin by the brushless motors, the overall system behaves like a bike or motorcycle instead of a monowheel. It seems to be a little janky right now but it seems to be fixable as well, I am really hyped about this, could you make one with a more common (easier to find and cheaper) parts version of this, the scale can change but again with gyro wheels? That would be great!
Im curious to see how this would work if you inverted it. The drive wheel on top. Would the cg be lower and maybe more stable.
will be nice to see it built with the gyro having an additional axis. They run the whole thing at some speed, as you did in the carpark with the previous version. Should be quite impressive.
A small little teny tiny suggestion James.
11:18 please please start adding panic button to your creations. You know the one button you push and it disables everything.
Such an elaborate floor polisher! :D
Thank you very much for teaching us about robotics.
When you tried the leaning steering, it behave like a bike. A bike must be first be countersterd to initiate a leaning and then steered as required. Maybe there is a similar thing for this?
13:03 Steve mold made a vid on bikes...
it has a simmilar effect on that too
Can you decouple the gyros from the rest of the drive mechanism? Very impressive though!
4:38
Nice idea, very innovative.
To solve the issue you described at the end of the video, would it be possible to have a mass at the top of the wheel move forwards and backwards to act as the forwards/backwards controls? This would allow the body to stay upright and avoid the gyro steering issue I think.
13:00 isn't this kinda like with a bike to turn left, you tilt right?
You can probably set those ESC in helicopter mode. That makes them first of all run at a constant RPM and it usually comes with ramp up curves
The offset center where it counter steers reminds me very much of how a bicycle front wheel steers, whenever a cyclist turns right, it first swings the steering bar slightly left before turning right.
Have you considered making a haptic chair like ready player one? Like maybe for vr aircraft.
It's not a failure, it's a successful demonstration of a way that doesn't work!
I would love to see a collaboration with you and the Stuff Made Here channel. Just think of the automated mayhem that would be unleashed!
you need two more gyros on the top and bottom within the wheel itself to counteract movement in the sideways direction that it falls over in. if you have 4 gyros you will be able to drive it.
As James mentioned, part of the problem is that the gyroscopes are leaning forward. I'd bet it would be less of a problem if it leaned backward. This is how bicycles' front wheels stay straight when you take your hands off the handle bars. :-)
I wonder if perpendicular gyroscope to monowheel could help with braking.
Wouldn't it be better if you place the accelerometer closer to the center of the robot? I've seen many projects where you place it in kinda random places.
That may impact the correctness of measurements and require compensations in the PID and control loop
Hey James i have fell of you're videos fore a couple of years.
you remember you're first big project. BB-8 v1.
This is wat wanted to do when you struggled with balance. 😁.
Lowe you're awesome engineering i want to learn to do stuf like you.
instead of changing the pressure angle on the spur gear, try increasing the backlash that way you have a wider area of contact to reduce wear on the gear while still having some extra space for flex in the gear
I was just waiting for a finger to get torn off and go flying across the room when you were trying to catch that thing. Cool robot though.
Seems like you should be able to use the turning angle to alter the gyroscope's center point so that when you "steer" it knows that the center of your angle is meant to be offset from your true vertical center. However, if you use the boat analogy, you should also be able to tilt your gyroscope physically if you place it on a horizontal rig.
could you could you make the gyro section tilt the opposite direction of the forward backward lean so that it would always be upright?
I'm wondering if the profile of the wheels that are causing issues with steering. When the robot leans slightly left the contact point moves quite far to the left, causing the actual balance to shift towards the right. Maybe try with a rounder profile for the wheels.
For the record, I know there are many liability concerns with a project like this. My newest wheelchair is from a small startup company. They were able to get around the regulations by simply marketing the device as a personal EV. If the item is not marketed towards the disabled, it is completely unregulated. This is how most of the smaller companies get their product to market
Wow - this is very complex to try it with the side-to-side balancing Gyro deflection axis' constrained to match the forward-back balance axis which, also is even further constrained to the forward-back drive force angle by virtue of balance-bots only having a single forward-back movement "joint".
I'm not smart enough to visualize how those PIDs would work together as a whole but, intuitively, it doesn't seem possible without decoupling at least one of those tied axis... It would be mechanically more complex to nest that Gyro unit inside an internally balancing ring to isolate the Gyro's vertical axis' but doing so would be dramatically simpler from a systems integration point-of-view since you could then initially work on tuning the bank and pitch PIDs separately.
If you also add an internally isolated Yaw Gyro pair, then you could steer it without roll/bank coordinated turns. Not likely your end goal. but once it can keep functioning and has a means of brute-force stumble recovery then there is the option of handing it off to a ML project to teach itself to coordinate the axis' and roll efficiently.
Super Fun Project and thought exercise!
Thanks for doing all the hard work!
Fantastic build 😊
I absolutely love it
Would it be possible to have a large flywheel to prevent a mono-wheel from flipping over on a hard break?
Putting the gyros in the wheel's center and putting them on their own race should fix a handful of issues. Centering them would disallow the tilting and turning induced by momentum, and putting them on their own race would keep the upright axis of the servo control. You also need some form of momentum dumping to desaturate the gyros anyways. If you really want a top-weighted wheel, to poorly emulate leverage force that a unicycle would see, I would suggest flipping your current configuration upside down, have the drive motor and batteries at the top, though most of this leverage will be made irrelevant by the heavy gyros in the center of the wheel; tilt to drive just isn't appropriate unless you have a significant lever on the robot.
Would it help to let the drive motor work through a differential and use a servo on the body of the diff for fine tuning while you're standing still? The motor is not agile enough for small compensation and causes overrunning the zero point. It does then introduce yet another mechanism...
Just my 2 cents
Hey James. Why are your gyroscopes not always centered on the robot?
You can build gyroscopes with bearings. Then it will always be in a horizontal position and can stabilize the position.
Why arent the gyro and the wheel just concentric? Wouldn’t that fix the problem stated at end of video?
What if the gyros had their own bering assembly so the robot can rotate around them when driving forward and the gyros can stay perpendicular to the ground.
I wonder what putting the gyro system on a limited gimble would do?