How this Active Gyroscope Balances

[@jamesbruton]

Thanks Morning Brew for my daily news perk - sign up for free here cen.yt/mbjamesbruton

[@elishagordon4852]

Cool Video Even Tho I Haven’t Watch It Yet

[@Dentalame]

A week ago !?!!??!!

[@lambsauce5445]

Can you do a ROS robot start to finish tutorial? The last one you did from your Nerf robot wasn't a tutorial per se, and it was a little confusing. If you do a sharp solid tutorial from 3d printing the parts, using the correct motors, and setting up ROS and talking to the Arduino and trying to make the system work (the most important part), I would love you even more - if that's even possible...More importantly, the ones online use the Turtlebot which is really pricey for everyone. So simply, please, please make a ROS robot tutorial showing how to setup ROS and letting it talk to the Arduino to make the motors move with encoders. If this is really confusing, please reply to this. But thank you!

[@balighgid6942]

OH YOU LITTLE, you made me buy mc Donald's with an ad for something else

[@jimmyhackers8980]

try dynamically variable PIDs. i guess you'd have to code it. i.e. the closer to centre/level the less aggressive the PIDs.

[@xitcix8360]

I know gyroscopic energy has been used for a while now, but it just seems like magic to see that semicircle just start tilting on it's own.

[@angst_]

Having the ball-bearings loose so they can roll around the track is decoupling them from the flywheel. The only thing that transfers their kinetic energy to the flywheel is friction.

[@jamesbruton]

Yep, a better design next week

[@CyclingMikey]

@@jamesbruton Cast a lead ring to put inside your gyro? Denser and thus heavier, and won't move.

[@pacman10182]

@@CyclingMikey something like woods metal can be cast directly in the plastic

[@StevenIngram]

I think the bearings are fine. The plastic parts just need to be printed to have holes for bearings and when you bolt the 2 halves together, it sandwiches them in place and holds them captive.

[@mikefochtman7164]

While that certainly affects how it comes up to speed (and you can hear them racing around inside when it stops), I don't think that in itself affects the gyro's moment of inertia. Once everything is up to speed, I think it works as well as a solid one would. (IMHO)

[@robertleroux5773]

You can use a car brake disc as a flywheel, they're cheap, come in various sizes and are already balanced.

[@jamesbruton]

Thanks for that. I also found flywheels for cross-trainers ;-)

[@elisonner5447]

That’s cool what made you come up with the idea

[@JoeyMcCart]

I wish I found this comment 5 months ago

[@thomb.9013]

@@JoeyMcCart why

[@Darenz-cg9zg]

@@thomb.9013 guess.

[@adissentingopinion848]

Aw man, I finally understand the mechanics of using a gyroscope as an active stabilizing component, rather than just a measuring tool. This is some choice educational content! I wish I could send this over to my younger self as encouragement for looking into physics more.

[@zetapi]

May the Forth be with you from Australia, James!

[@ShadowPhenix273]

Oh, why hello there fellow aussie

[@benjaminrogers9848]

@@ShadowPhenix273 Oh, why hello there fellow Aussie

[@ShadowPhenix273]

@@benjaminrogers9848 oh, why hello there fellow aussie

[@kingmasterlord]

hope you were prepared for the revenge of the fifth!

[@Idontknow-km5hp]

HOW DARE I FORGET THAT IT WAS FOURTH OF MAY

[@sean63b]

pid loop tuning tip: start with i and d set to zero. increase p slowly till it starts oscillating, then try to increase the d gain to dampen the oscillations. Go back to the p gain and resume increasing. do this until you are unable to dampen the oscillations, then lower back to the last controllable value. For this particular case, the i gain should be unimportant, since that just ramps up slowly in case the p gain gets too low to drive the motor to the exact setpoint, which shouldn't be a problem for a balancing robot of this type since it's reacting quickly.

[@graealex]

11:50 A gyroscopic roll amplifier for ships would be something new.

[@blahorgaslisk7763]

Could be used in training vessels to simulate rough sea. Get those landlubbers green and puking their guts out without having to wait for a storm... Can't help but wonder if reaction wheels might not be better for something like this though.

[@graealex]

@@blahorgaslisk7763 Or just to flex on other boat owners. Imagine your boat sitting in the calmest waters ever, yet it's rolling like you're in the mother of all storms there ever were.

[@daleguenther5826]

Thank you for creating another great vlog! I can always depend on learning some thing from your videos!

[@fraserbuilds]

I love this style of video of demonstrating specific concepts with robots. Its so cool and rlly inspiring.

[@OrionAerospace]

We are putting a reaction wheel on our rocket to control the roll! Awesome video as always :D

[@stochasticsignal1951]

Always funny running into KZbinrs in the comments. Where's the next Kalman filtering video? ;)

[@hughclinton]

Reaction wheel or a Gyro. I'd have thought a gyro would make more sense?

[@mikefochtman7164]

Great! You've combined two of my favorite subjects, gyros and PID controllers. While in the Navy, came across a test hull system that used two huge gyros to force the hull to rock side to side at different rates in order to test some internal systems. The rate of roll was just a matter of how fast they would pitch the gyro's fore-aft. (of course, those gyros were about ten foot across and really massive)

[@fatalmystic]

Finetuning a PID control is an art! Nice job! :)

[@Andreluisdias23]

13:22 you noticed that the board to which the accelerometer is attached, is vibrating a lot, maybe with a little adjustment the device will be even more stable. great video

[@BlueScreenCorp]

I appreciate how interesting and informative these videos are, I have never seen the inside of a gimble before and now see how it responds to movement makes it much more intuitive

[@benjaminrogers9848]

There's a Tom Stanton video on drone COG that explains why adding weight to the bottom made the balancing worse, broke my brain tbh

[@blahorgaslisk7763]

Thought of that, or perhaps a similar video, when he added those nuts. I still can't wrap my brain around exactly how it works, but when I watched the video it was explained so well it all seemed perfectly logical, which I expect it is. But now I can't remember the details and it sounds totally weird again. I remember it being explained how adding weight *above* the point of balance was much safer. Also how it got easier the further above the balance point it was. Kind of like how it's easier to balance a three yard stick on your finger than a foot long stick with the same weight. All in all it hurt my brain... Oh no! It's coming back to me. The problem with using a weight at the bottom to balance a helicopter is that as the craft changes direction the weight will act as a pendulum. Go forward and the weight will make the craft tilt forward. The more the craft tilts the more power is projected backwards accelerating forward which will make the pendulum force the craft to tilt even more and so on. Same with air resistance which will cause the craft to tilt in the direction it is moving or if it's wind then it will tilt towards the wind. All in all it becomes very unstable. With the weight above the craft will tend to tilt in the opposite direction, effectively counteracting the tilting effect of the acceleration. And so the craft will be more stable but less maneuverable. Ouch! Brain not happy that it actually had to work for a few minutes there... Now I can't immediately visualize how this would influence a gyroscope. Late night here and my brain says it needs it's rest.

[@anactualpilot]

Been watching for a couple months now and your uploads are always so well made! Great content, thank you.

[@toasty8432]

You have a lot of trust in the strength your printed parts, standing chest level to what could been thought as a fast spinning ballbearing bomb. I wonder what sort of velocity one of those would come flying out at if the plastic bonds gave way?

[@jamesbruton]

Not much, they are only 160kv motors

[@daniellaroche8690]

The word for this in my mind is "centrifugal claymore" 😹 (sick robotics though! This is extremely cool)

[@Nickgowans]

@@jamesbruton that doesn't really matter though, what matters is the RPM, the diameter of the wheel and the weight of the ball bearings. You should fairly easily be able to calculate the launch velocity and energy if it should fail

[@conorstewart2214]

@@Nickgowans it would be quite complex to accurately calculate it, if not impossible. Sure you could handle it as if the casing just suddenly didn’t exist but for the ball bearings to come out the case would have to break apart and that would probably take a lot of the energy out of it.

[@PhaTs00p]

That's a lot of confidence placed in the strength of the 3D parts.

[@paralexvr]

This is incredibly unsafe, if the plastic cracks you get high velocity metal ball shrapnel in all directions... everybody dies in the room.

[@jamesbruton]

It's really not going that fast - they are only 160KV motors.

[@PhaTs00p]

@@jamesbruton I'm just pulling numbers out of my ass. If I understand electric motor ratings correctly that's 1900rpm at 12V, at 60V up to 9600rpm. Ball bearings ~9cm from middle results in a RCF (g force) from 360 g (1900rpm) up to 9200 g (9600 rpm). Even if you do trust the plastic I strongly recommend balancing the wheel properly. Additional fun facts, assuming a weight of 4grams for one ball, the kinetic energy at 1900 rpm is around 7.5 Joules per ball, exactly the threshold muzzle energy you'd need a firearms license in Germany if you planned on launching the balls. At 9600 rpm we're looking at 180 J, about the same as a .32 S&W Long cartridge from 1896. Modern 9 mm ammo is around 550 J.

[@ramakrishnanr3877]

I have stumbled upon an article about an old russian self balancing car.The inventor used pendulum as a feedback.The pendulum is connected to a rack and pinion which changes the angle of the gyro when tilted.

[@jamesbruton]

That would probably work, but it would be hard to tune ;-)

[@BirnieMac1]

Have really been loving your videos for learning more about these kind of engineered stabilisation systems

[@rpavlik1]

Really cool! Thanks for clearing up the difference between reaction wheels and control moment gyros too.

[@northhamptonshire]

I love this ongoing series on separate mechanisms!!

[@jamesbruton]

Thanks, more coming soon!

[@ImperialLightandMagic]

Anyone else thinking the loose ball bearings spinning at that rpm was creating a 3d printed claymore? :) Joking aside, amazing work as always though James

[@wrxwalsh]

I had reasonable confidence in the material, but that was my first gut thought

[@RandomPerson494-12c]

I have been waiting for this. Thank you James !

[@divyaprakashbiswas8781]

Your projects always motivate me. Thanks for sharing.

[@3DPrinterAcademy]

great video! really enjoyed it!

[@Avetho]

This reminds me of when I spun up an ancient 8MB, 8 platter HDD, it was huge and I could very clearly feel the gyroscopic forces as I tilted the drive. That was cool.

[@blahorgaslisk7763]

There was a reason those huge 8" and larger drives were so heavy and in some cases even bolted to the floor... The largest (physically) drive I saw had a aluminum drive enclosure that was about a yard across. There were a drive axle sticking out with a freaking pulley on it and a belt connected that to the drive motor. A belt driven hard drive! It seems so insane now. And I think it was something like 30 MB of storage! But hey, it replaced a few of those 8 disk drive stacks that they were throwing out. I think those stored somewhere around 450 to 800 KB each. I may be way off, but this was around 84, perhaps earlier. Can't remember exactly. That "new" hard drive was not only way higher capacity than the drive stacks it replaced but also very much faster and more reliable. I did work on some 8" drives later, say some time between 86 and 90 or so, and those were about 30 MB and on their way out then. It was kind of hilarious. You had a tower computer with a controller card that connected it to a second tower that was almost exactly the same size as the computer and that held one 8" HDD that could store about 30 MB. I'm trying to remember the brand but it's hard. Priam keeps popping up but it doesn't feel quite right. Googled and yes it probably was Priam. Damn brain, you did good!

[@Markste-in]

A good alternative for the balance wheel would be a car disk brake. They are heavy, can be super cheaper and are fairly centered and balanced

[@Diamonddrake]

This is pretty cool! I didn’t know about that seakeeper, what a neat idea

[@Callofdootie]

I watch and love all your videos. This one for some reason seemed really satisfying to watch.

[@MrRitzyVlogs]

I really liked seeing the acceleration plots at 10:40. Would have been interesting to see the plots for the closed loop tests.

[@gingermanc]

Thanks for inspiring kids ( and adults everywhere ). You are machine ( in that you output a lot of high quality stuff ).

[@TS-kg4lf]

Nice project to show and tell how to tune PID on one axis and what it is used for :)

[@jannic7506]

Your work is awesome. For me it's impossible how laborious and fast you can make videos🔥 P.S.: Thank you so much for sharing your experience and experiments. I always learn something new on your channel

[@jamesbruton]

Thanks!

[@Danucart]

Sorry, I commented on the disadvantageous attachment of your IMU earlier, while still watching your video. You saw it for yourself, well done! Looking forward to the next one!

[@jamesbruton]

More next week, thanks!

[@thesoupin8or673]

Incredible. I've enjoyed your content for a little while, but this one earned you a sub. Great content!

[@jamesbruton]

Thanks, more next week

[@josiahmitchell530]

Good job with all the ideas. I'm enjoying them and its cool learning about new concepts! Keep it up!

[@reggiep75]

15:45 - I had wondered when the compensating pairs would arrive and even tho you need a lot of space they do fix issues.

[@redbaronrefining5322]

Ah, inertial gyroscopes. I have a few of these in my collection. Amazing guidance technology back in the day! Talk about some killer bearings in them!

[@ReiverBlue1971]

On the "keel" test it looked like a bit of a "Tah-Dah!" (rather please with yourself) moment ;D Wonderful work

[@_BangDroid_]

15:55 That's the cutest goofiest robot I've seen in a while 😆

[@jamesbruton]

It's good isn't it

[@rcleveacp]

Wow! Very impressive. Keep up the great work and thank you.

[@514_cricket]

YASSS the 3d printing music is back! Thank you!

[@AmaroqStarwind]

You should make a design that has a gömböc shape in it.

[@johnflux1]

Instead of PID, have a look at Model Predictive Control. It is a bit more complicated, but the result doesn't need tuning and the result is optimal. For gyro control, it is perfect, because it has a very well known and simple physics model.

[@jonmendenhall5074]

Yeah definitely. Slapping a PID on everything might work, but if you know model... you can get really solid performance like you said. In this case he probably would want to monitor the angular acceleration of the hull, then exert an opposing force on the hull by exerting a torque on the gyro through the servo. And then have an outer control loop (much more acceptable to be PID) that monitors the angle of the hull and feeds into the desired force for the inner loop.

[@zarro110]

Just print little notches in the outside of the ring where the balls are. If you put as many notches as balls they will be pushed against them with the centrifugal force and space out evenly. Also makes for better power transfer to the balls without needing much precision

[@jamesbruton]

yep - that's happening in V2 next week!

[@literate-aside]

This is absolutely fantastic

[@locostse7en]

You have made me fascinated with all this.............

[@FeartheLess]

awesome video as always! I always learn something new haha. i have a question relayed to gearboxes, have you considered 3d printing a planetary gearbox? if so why or why not, are you planning on making a video about them in the future or are there any limitations to printing them? :) you’ve also inspired me to make my own cycloidial gearbox for a 28byj, it currently has 1 disk and 45 decrees of backlash, hopefully i’ll improve it over time haha thanks for the amazing work and videos!

[@davenarisotto3674]

Hey james, very interesting video as always. Do you have any plans on doing anything related to mecanum wheels?

[@jamesbruton]

Yep, kind of, in a couple of weeks!

[@madimadisofian]

Hey, if you‘re imu is producing very noisy data(for example because of vibrations) , a kalman Filter could help out. It can be implemented in the software 👍

[@pespsisipper]

7:30 this is incredibly unsafe! how can you stand so close and just look at it?!

[@hakimchannel]

wow, nice job friend

[@jasonhendricks4802]

Make a silicon mold of the track the bearings ride in. Cast the silicon in plaster and then fill with lead. Making a solid ring will improve the stability.

[@ivan_959]

You could stack hard drive platters to make a flywheel and spin up the hhd motor up. Or stack platters and use the motor hub only.

[@matthewwillshee203]

Interesting stuff, although I have to agree I am a bit worried about that rotor bursting. I think the control would work better if the output of the PID controller set the servo rotation speed, rather than angle. The torque reaction from the gyro is proportional to the rotation speed of the gimbal. Then, to keep the gimbal position, from tending to infinity, an outer loop that reads the gimbal position and alters the target lean angle.

[@sebincjose9474]

Actually I was thinking about this from a long time. Placing two opposing gyros on the hip of a humanoid robot and balancing them. Hop you try to make one.

[@sill2323]

Would you mind trying the first gyroscope gimbal with the weights on the top of the red frame? To my eye, the balancing bicycle might have this arrangement?

[@JulianMakes]

I can see a metal lathe in your future James! I got a lovely boxford (old British make). Imagine the energy in your gyro x 10 :)

[@conorstewart2214]

You have essentially stuck that IMU out on a springboard, you can see how much it is vibrating, that will probably affect the roll calculations and add a lot of noise and hence affect the whole control system. The IMU should have been mounted to the main body of it at the very least, maybe with some light vibration damping too like they do on drones.

[@AlexP-zx1yo]

At the 0:50 mark you can see the gyro's base start to move in one direction as the disc starts slowing down. Is there a name for this phenomenon? How is the rotation transformed into translation in this case?

[@1FishinAddict]

This is great for a kayak stabilizer !

[@jordan1734]

You don’t disappoint. 👍

[@sreynoldshaertle]

Try a markov decision process! A good enough policy will actively combat windup, which is pretty neat. Maybe more useful for reaction wheels than gyros, though.

[@oisiaa]

I think James has printed more with a 3d printer than any other person on Earth.

[@eongan]

I think mechanicaly isolating the gyro from the system would improve signal to noise ratio, thus you can get much higher control bandwidth without causing oscillations.

[@Nickgowans]

This is terrifying especially used on a bench as waist height. If the flywheel fails you will have dozens of pieces of extremely high speed shrapnel flying at your stomach. It's why people wear face protection when using angle grinders, not because of the sparks but in case the grinder fails.

[@Genubath1]

I would recommend working behind a screen so that if your fly wheel breaks apart, you don't catch high speed ball bearing in the face.

[@RixtronixLAB]

Nice video, thanks for sharing it :)

[@wilmantube]

Could you explain your implemenation of PID in Arduino code & motor control a little more in depth? Amazing vids as always James!

[@tannernewton8543]

09:02 Did you mean, Compensate for the weight of the servo?

[@SpeedrunnerG55]

this is just like the two-wheeled electric car designed by lit motors. counter-rotating counter-processing reaction wheels for stability control are also used in the Hubble space telescope to keep precise track of its view. however, rotation of the counterweight becomes ineffective after the axis or rotation aligns with the axis of the vehicle rotation a secondary desaturation force is then needed to realign the wheels to take the rotational energy out of the system. satellites such as bubble use thrusters or magnetometers to achieve this task. thus while on earth, it advantageous to change its stance such that gravity itself acts as the reaction force to remain stable against any outside force.

[@adrielparsons7790]

Hi James liked your build were did u say you were publishing the code i would like to have a go at building one myself from billet alloy or milling machine for my boat any help would be much appreciated thanks

[@MaxWattage]

You have a plastic container with anisotropic weaknesses due to being 3D printed, filled with heavy ball-bearings, and then spun at high RPM to maximise their kinetic energy. Congratulations, you have made a Claymore anti-personnel mine! I really wouldn't want to be in the same room as that.

[@DoRC]

9:30 I wonder how much damage all those steel balls would do if that disc decided to fly apart

[@Clovenlife]

Would love to see a breakdown on how you print functional parts that survive high stresses

[@zippythinginvention]

I'm very excited about the trajectory of this. Have you given any thought to unidirectional force?

[@jamesbruton]

Next week's video is two gyros

[@zippythinginvention]

I've got a lot of insight into this, if we could have a private discussion at some point.

[@stochasticsignal1951]

I've been working on a CMG for a while. I always like seeing other implementations. Does the vibration have a noticeable negative effect on the IMU's readings? Are you using any filtering methodologies?

[@jamesbruton]

Possibly, and no ;-)

[@stochasticsignal1951]

@@jamesbruton Thanks James. Amazing what you can do with a $10 IMU today.

[@tiredowl2687]

You should check out "UAV Tech" channel about tuning the PID loop. His video is about drones, but the PID tuning process could work well for you as well. He also has a PDF explaining the process as well. The PDF is very detailed and I like it more than the videos.

[@e0da]

Hi James! I worry that if something lets loose all that kinetic energy in those ball bearings could do a lot of damage, especially to squishy organic stuff. 😬 really cool project! But spinning stuff can be so dangerous. Take care! ❤️

[@Nickgowans]

This is my thoughts too. If the 3D print fails then you essentially have a nailbomb launching high speed ball bearings around the room.

[@rmdavidov]

I am early and already know that this will be a great video

[@kadenwerner7681]

Hello, I have been trying to create a oscillation-control system like this for a school research project. I have managed to setup / tune the PID control for this fairly well, however I'm running into an issue where the flywheel cradle can't return to zero without creating torque on the hull. I'm already running my flywheel's drive motor at it's lowest speed, and was wondering if you had found a solution to this in this project.

[@azyfloof]

I think the biggest problem is how sluggish that servo is. You can see how much it lags when you first demo it with the potentiometer. Try one of your cycloidal gearboxes and a brushless motor for a more snappy and responsive setup :)

[@Gunstick]

Would a reaction wheel be lighter than a gyro for stabilizing a walking robot?

[@jamesbruton]

Difficult to say. Also the reaction wheel would still have gyroscopic precession when it's moving which would do some funny things to the stability in a perpendicular axis.

[@Gunstick]

@@jamesbruton maybe that's the reason space probes are stabilised by gyros and rotated by reaction wheels.

[@RexusKing]

Did you use high pass filter to kill the IMU noise mounted basically on a swimming pool springboard?

[@jamesbruton]

Nope, the IMU has it's own internal DMP though - I didn't look at the output when it was spinning tbh

[@RexusKing]

@@jamesbruton Cool, the results is good nevertheless.

[@johnflux1]

Can you please add links to the description for those youtubers that you referenced?

[@jamesbruton]

Yep, done!

[@RupertBruce]

@3:20 why? Gyroscope creep is inversely proportional to gravitational force? Odd...

[@garnergc]

It would be awesome to see a MPC controller implemented for something like that. But short of having a FPGA to do the computation there just isn’t enough power in a microcontroller

[@gyrogearloose1345]

Thanks for this. Valuable intro to gyros ( my gyro!) and control systems with the dreaded (and beloved) PID loop. Very stimulating. Question: what is your servo tick rate for the PID? I'm amazed that little Arduino can even attempt it.

[@jamesbruton]

PID is running at 100hz, but the IMU data is only read at 25hz :-)

[@gyrogearloose1345]

@@jamesbruton Aha, a bit slow isn't it? Probably first thing for better performance will be faster PID and IMU. Maybe Teensy 700MHz and .... ?

[@markhorstmeier8734]

Would a stepper motor that allowed free rotation until it was energized allow you to remove the torque component from the servo axis? Not sure if the startup time would be too laggy

[@sddiymakeitworthit7512]

Always awesome 👏

[@kylebrake1806]

I love your videos!

[@GEARSinc]

Can I ask what 3D printer you prefer for these projects?

[@fahmiyusoff9154]

If the fly wheel is heavier it will Be Moré efficient right?

[@Xlaxsauce]

Is there a practical way to measure the input force and acceleration? Then they can be used to modify the response of the PID per "input". A problem I see with my suggestion is whether or not the board is fast enough to respond to the intial motion before it tilts too far.

[@soweliLuna]

it seems like you had the PID control the position of the gyro, but the gyro only applies a force when its pivoting. in the next version, you should have the PID control the pivot speed rather than its position, if you haven't done that in its current configuration, the integral probably acts more like the proportional is supposed to act

[@jamesbruton]

There's another version coming next week with an observation controller, kind of ;-)

[@firstnamethenalastname]

16:08 those graphs look like they represent the pitch of your voice during one of your videos i'm not sure how you'd build something to stabilize it

[@Pfaeff]

Now I want one of these toy gyros. Is that one just one of the cheap ones you can get on amazon?

[@jamesbruton]

yes

[@narwodev]

Wahooo, it's Tuesday :)

[@CaptM44]

you should do a video on magnetic gearing