How Many Ways Can This Robot Steer?

[@nathanskinner423]

Now you need to code a more advanced version. The ones in FRC do a vector addition problem. It allows you to request a position and request a rotation with command inputs and the controller will figure out what direction all the wheels need to be turned and what power to give them. Usually 100%. It is pretty crazy some of the stuff you can do. with this method it is possible for the robot frame to be spinning and driving in a linear path.

[@comedy6631]

Our frc team actually has field oriented code, kinda unrefined though, we are working on it.

[@tirthjain1975]

Great idea 🙌

[@CuulX]

How would that look and work with this kind of robot? Are there robot videos on YT that show this mechanical setup (no omni wheels) being able to rotate and drive independently of the path curve?

[@nathanskinner423]

@@CuulX FRC robots have mechanisms where the wheels can rotate infinitely. So there are no issues with running into a servo limit.

[@alexandregrynagier1762]

@@CuulX You would ba able to move as in a fps game, and if using fpv, you would get similar sensations. Probably not the most usefull think on earth, but for example if you had to use a forklift vehicle, these controls could be very usefull.

[@Paul-rs4gd]

This drive system is way more effective than I expected, and the design is commendably simple. It would also be interesting to see a differential swerve drive - there are a lot of interesting ideas out there. It could use bevel gears, or it could use a slipring.

[@AmaroqStarwind]

Mmm. Triple-Differential~

[@rohanbandaru]

High-school robotics teams have been using this type of drivebase for years. However, one problem with swerve--drive is that there is a bit of a delay when you want to change direction, as you have to wait for the wheels to rotate to the desired directions. With Omnis or Mecanums the change in direction can be almost instantaneous. Anyway, for some next steps, you should consider improving the algorithm for turning the wheels such that you can input the desired translational and rotational velocity of the robot. This will allow for true holonomic movement.

[@noahofdryveaerospace805]

If you can, and assuming you are using SDS modules, use Falcons rather than neos for steering. Our team has experience with both an all-falcon drive and a falcon (driving and NEO steering). It would be neat to see a Swerve Drive go-kart that uses a joystick, and would be fun to rip around on, especially now that SDS sells a Colson Wheel Kit.

[@WeAreTheTrueMedia]

That's the most useful omnidirectional I've seen. True direction change that doesn't need a perfect surface.

[@christianstorms3950]

Plus it can run on inexpensive run of the mill tires

[@UCs6ktlulE5BEeb3vBBOu6DQ]

You are the man that really should own a plastic crusher/melter filament recycling machine with all that printing you do !

[@lio1234234]

Do you know how expensive and difficult it is to process your own plastic back into filament?!

[@sajalagarwal195]

@@lio1234234 well not really. I made it for my college project and I was able to get it done for under $400

[@cx765]

@@sajalagarwal195 What was your filament diameter variation? And was it consistent across batches?

[@lio1234234]

@@cx765 precisely! It's impossible to get perfect prints when you have flow differences due to varying diameter of the filament!

[@bo0tzz]

@@lio1234234 You can compensate for this with a filament thickness sensor on your printer though.

[@Xlaxsauce]

Wow, that is really responsive. Seems like a simpler method rather than those expensive niche wheel types. The design seems sturdy and reliable

[@Gee-Dee-q1e]

You've once again made a case for why I need a 3D printer. I've been working on a 4WS vehicle for a while now, but without custom fabrication available, I'm still left to figure out how to rotate the e-scooter wheels I'm using. Your video made it "look" so easy. Cheers! Now I have an example to follow for my own project.

[@beautifulsmall]

Thats a really smooth result , great ground clearance, omindirectional with full grip. Nice work.

[@blinded6502]

I recommend learning geometric algebra, it simplifies tons and tons of difficult linear equations to a single tiny human-readable equation or two. It treats rotations and translations as reflections, which allows a lot of terms to cancel out. This lets you come up with transformations, that are far better than quaternions, and that allow you to easily mix translation and rotation. In fact, quaternion - is just a bireflection in 3d. Bireflections in general sense are called rotors, which actually work for any amount of dimensions. I could honestly talk about beauty and simplicity of GA for hours. You have to learn it, at least for times when you get stuck with mathematical derivations. It would allow you to program a lot more difficult robots with little thought. You could easily make an inverse-kinematics algorithm, that works with various motors, pistons and hinges of all sorts of degrees of freedom. It might be hard for someone unfamiliar with terminology to get into GA, but believe me, it is 100% worth it. The actual math is far simpler than you might be led to believe.

[@doctorbobstone]

Is geometric algebra something akin to a branch of mathematics? Or is it a more focused set of ideas than that? Also, where would you point people for an introduction?

[@blinded6502]

@@doctorbobstone There is a community driven website about geometric algebra, under the name of "Bivector" but youtube keeps on removing my comment with a direct link

[@pws354_8]

What math prerequisites are required to be properly prepared to study such material? Differential equations? Linear Algebra?

[@blinded6502]

@@pws354_8 Mostly just good intuition about linear algebra, I'd say.

[@blinded6502]

@@pws354_8 Knowing group theory would also help grasping some core concepts.

[@sushsidnd]

I thought I knew how swerve steering worked before watching this but wow I was wrong. Very educational video.

[@richhandler2621]

right? I really felt humbled lol

[@henricoderre]

Very nice mechanism. I love that you can turn instantly in a different direction.

[@DeanBrettle]

Consider replacing the current wheels with omni wheels and using the ability to control the angle between wheels and the current direction of travel to create what amounts to an infinitely variable transmission. Imagine the wheels initially pointing in the direction of travel and then gradually pointing more and more inward as speed increases. Would be really cool to see how fast it could go.

[@stoatystoat174]

moves hypnotically, loads of intresting stuff in twelve mins, good diagrams and like showing the wiki page for further reading

[@Variety_Pack]

This would be an excellent platform for a custom built slipring! This would let you rotate the wheels as many times around as you want, although you'd need encoded motors to drive the rings.

[@rossfuller3224]

He is using encoded motors

[@ericlewis3444]

Man, I look forward to these videos every week. Genius stuff

[@justinberdell7517]

Of the omni robots I've seen on this channel this is the best

[@mortgeek]

Swerve has gone mainstream

[@AmaroqStarwind]

If you use a less aggressive steering angle in the "rear" wheels, you can keep moving forward throughout a turn. Also if you steer all four wheels in the same direction at higher speeds, you can get more stability. Anti-Ackermann arrangements help out here too, and having differential steering on top of that is always fun too.

[@TheMeditron]

I really like the way it moves, it feels satisfying to watch. could see it being really useful as wear house equipment or maybe a robot to deliver food in a busy restraunt?

[@TheMeditron]

@@kruleworld oh ya, for sure! anything that needs precision movement would be super handy

[@ShadowRam242]

@@kruleworld Already done, I made a hydraulic lift that drives exactly like this in 2016

[@real-1982]

I would love to see you doing courses that are more detailed and shows all the steps, I would pay money to learn it from you.

[@c.kur013]

Same

[@SuperBrainAK]

that looks cool! I like the added together steering. It can truly go anywhere!

[@WFCinSC]

I would love if there were 4 independent remotes (one per wheel). Then teams of 4 could compete by trying to complete some simple task in the minimum amount of time. It would be a great demonstration of how inept human communication is for complex tasks, how inept humans are at translating verbal commands to actions and not a bad team building exercise to boot. I wonder if we could beat an AI...

[@ledocteur7701]

beat an AI ? absolutely not, even an AI running on just an Arduino could beat us at something that "simple".

[@goplayman5658]

Hi James! All your projects are always top notch! Could you make a video on how you plan your projects? From idea conception to CAD design to 3D printing to wiring electronics and all that. It would really be helpful to upcoming Robotics Engineers like myself. Thanks in advance!

[@surfcello]

I think if the wheels could swivel around infinitely often, it would be cool to decouple the movement direction and the orientation completely in code. Then you could move around like in an FPS video game, with one joystick controlling the motion and the other controlling the orientation. This might be useful for battle bots.

[@Lucien86]

Nice. One of my favorite robot / car configurations.. Great Vid. 👍

[@scootergem]

You've really come a long way and it's interesting to see what you're up to next. Keep up the good work and thanks for sharing.

[@ianjensen3890]

Fun to see a video where other people are mentioning FRC and the popular Swerve drive!

[@josiahmitchell530]

That's a really cool concept! Great job!

[@EvileDik]

Amazing, that looks really satisfying to drive! I'd love to see more ROS/ROS2 stuff (especially openVslam), remote control 'robots' are all well and good, but aren't going to be overlording anything anytime soon!

[@lucywucyyy]

you should make the centre of the turning circle move closer till it goes to the centre of the robot and it rotates on the spot, this is how ive built 4 wheel steer robots like this in games, it makes a very intuitive and fluid transition from straight to turning to rotating and back again

[@lmao2678]

There are some high school robotics teams that build something like this but with infinite rotations for steering, you can see it in action here. kzbin.info/www/bejne/rXTEnK2Dm7eDgdU

[@willowmacivor6931]

SWERVE DRIVE GO BRRRRR

[@Goodgu3963]

Damn their media team knows what they're doing! I have watched a LOT of FRC videos in my years and they've got some of the best videos I've seen.

[@matthewthomas9383]

Let's go team orbit

[@grey1185]

Omg! This is the same design as my rover! This video will greatly help me improve the chasis and control method

[@randomelectronicsanddispla1765]

Probably my pick out of all your mobile frame designs. My favourite design is still the six wheels and rocker-bogie suspension used on Sojourner, Spirit, Opportunity,...

[@DktheWelder]

Wow a perfect base for a robot! Very cool

[@samuzamu]

Simple yet effective design. Seems really nimble and quick. One thing I noticed, though, is that it seems to speed up in curves. Do you run the inner pair of wheels at the same speed? If they slowed down slightly when turning, and you modeled the line of turning between the inner and outer pair of wheels, I think you'd get consistent speed.

[@jimbarchuk]

That sounds correct. It seems that he set the inner wheels as 'the speed' and pushed the outer wheels faster too keep up. It works but would be unusual and uncomfortable as a passenger. The 'outer side passengers' would be constantly thrown against the wall.

[@scrapwomblecreatives6944]

i love these shows, them servos are to die for

[@dalerogers1134]

Outstanding! A future Mars rover will also use the omnidirectional system you built. I have developed an omnidirectional standing wheelchair prototype that uses three Omni wheels, but I think it would do better with the drive you built. That is not something I can do myself, but the future, as always, is unknown.

[@PeTr01]

Have you thought about using a component like cable down the wheel shaft so you can power the motors without having to worry about wires restricting your motion. I’d assume the motors can overcome the friction caused by the cable connection (though from my experience with CRT’s they can get loose pretty easily) but you’d get 360 degrees of free XY plane (floor) rotation

[@davidmalkowski7850]

An easier component to use for that task is called a slip ring!

[@PeTr01]

@@davidmalkowski7850 I didn’t know about that! So cool!

[@killapoof]

FRC does something even easier, we put both motors at the top and drive the wheel with a bevel gear at the center of rotation, look up SDS mk 4 for a good example of this

[@PeTr01]

@@killapoof that’s so clever! I’d imagine the wear on the gears is less than possibility of dust/buildup affecting the electrical and magnetic signals of the previous two suggestions and it keeps both motors stationary as well rather than a rotating motor (not in the way that it should) creating unnecessary toque

[@dscrive]

This reminds me of the Jeep Hurricane concept vehicle built a number of years ago. I was just a teen when I read about it, but boy did I want a jeep that could crab "walk" into a parallel parking spot or spin in place.

[@BenKDesigns]

Thoughts for a next video: Make it full-size, or at least human rideable. Maybe get ahold of Mr. Furze. I can't imagine the crazy things you could do with a car that could drive like this, but it'd be fun to see. :D

[@msm2029]

Great Job! Loved the project. Perhaps one thing that could be done is to add some gear ratios to make the wheel turn 360 degrees while the motor does half that.

[@martindinner3621]

On the servos, or the wheel motors?

[@c.fisherfighter638]

I find this this project that seems '' simple ''. Is actually pritty cool

[@countZ74]

2:1 gears on the steering for 360° rotation and to offset the steering axis from the servo axis. Then you can use a hollow axle to run the motor wires through. Or stepper motors/CR servos and an encoder for multiple turns (limited by the motor wires twisting).

[@dubov_ski]

good ideas, but 360 doesn't add much functionality, since you can already reverse the voltage going to drive motors and cover the remaining 180 degrees with the current setup. turning would be twice as fast though (and half as torque-y).

[@ShaunCullen]

Mr. Bruton is playing Banjo Kazooie Nuts and Bolts in real life and I love it!

[@mattiasfagerlund]

That's awesome, I'd love to see a version you can ride...

[@Manabender]

The wheel mounts definitely look like the weakest structural piece, you would need to make those significantly sturdier if you wanted to support the weight of a person. Not to mention the weight of the rest of the vehicle. Probably possible, but not sure about practical.

[@mattiasfagerlund]

@@Manabender But imagine him riding around on a chair like that, at a convention, that'd be so funny. Using 3D printing alone would probably be hard, but he's got a CNC and a welder...

[@LucGendrot]

This would be fun in combination with the Really Useful Robot!

[@spykillergames8402]

Ya that base combined with an arm and some more complicated computing could do things like opening a cupboard door etc

[@Fungusamnguss]

I'd love to see a bunch of these running some synchronized kinda of dance

[@vinyl2780]

Swerve drives are so cool, I know I a lot of people have probably commented this, but take a look at some FRC bots that do this :)

[@andrewslater6846]

This would be interesting to develop an RC race car that has 4 individual steerable wheels to optimize steering angles to go around turns on a track. And then analyzing how the center of gravity effects how the front/rear steering should be managed.

[@n1elkyfan]

One big improvement I could see is having wider wheels that the motors can fit inside of.

[@amazinghistoryofvlogging6894]

Great video, really interesting. Keep up the great work, you are an inspiration to so many.

[@jamesbruton]

Thanks!

[@Fireheart318]

Have you ever seen the 2008 movie, Speed Racer? Some of the cars in the movie, called T-180s, use a jet engine in combination with that type of omnidirectional wheel to vector the engine's thrust in any direction without losing wheel grip or requiring (even more) insane engineering to create a nozzle capable of vectoring a full 360 degrees.

[@ThePyrojen]

love all your projects! keep it up!

[@Saraseeksthompson0211]

Been looking forward to this. Good job as usual!

[@alexmcconnell6637]

Now all you gotta make is a differential swerve drive, that shits wild

[@jakobjorgensen7773]

Brilliant steering capabilities!

[@Ritefita]

3:23 With offset on wheels you could change "robo's width" and shift support under CG if needed. could have different stances

[@iAmTheSquidThing]

This is like one of those remote platforms they use for moving huge parts on construction sites.

[@medienmond]

This is insane. I want a platform like this for my next robot...

[@dylanw.4533]

A platform like this that could be given a route and ideal speed, and decide which method of steering allows for maximum speed around a course a normal RC car couldn’t even handle.

[@PenuDjira]

Love From Indonesian For You Mr. James🇮🇩

[@crapadopalese]

To me, the most appealing solution here is not a mechanical one but a computation one. Have a computer program re-stabilize the wheels on their big axis after performing a turn. I wonder how much time would the wheels spend not on their big axis in this case...

[@conorstewart2214]

You could also try different designs of omni wheels, you get ones where the wheel is just many diagonal cylinders around it and it should mean you get the same contact area in every direction.

[@ryanagamalian9892]

My robotics team has made a really creative fix for omi directional wheels it’s called swerve drive

[@czechjuli94]

Asome pod for cinematic shots

[@kurtiskill2847]

It would be sick to have a go kart that drove like this.

[@jed3255]

kzbin.info/www/bejne/m3aTfKqkeb6Mnpo

[@Acaykath]

Maybe add automatic vertical leveling to deal with rough terrain? Of course for city driving, you'd have no problem getting into or out of a parking space.

[@jackalovski1]

if you program the outside wheels to spin a little fraster than you already have them, you have then got active yaw control.

[@Redneckmfg]

I wish I had 1/100th the knowledge you have.. your incredible....

[@dibrockly]

Great build!

[@heydeohgee]

Extremely elegant! so cool

[@lonewarrior6633]

Happy Tuesday James!

[@theob1712]

If you could motorize that ball wheel and slap a few under a shoe, you could probably make some sort of vr treadmill.

[@whetan]

All the FRC kids gotta be going crazy seeing this

[@freyabauer8662]

we are going nuts as we see him doing swerve legit love swerve and how powerful it can be done with enough knowlage

[@CalloutWraith]

I really really like the Idea behind this. As a Project its great, to learn the Stuff in Robotics. As Ing-B, my thoughts was first like - "mhhh in an industrial Environment - what is about the wear/tear on specific Points on the Wheel - is it cheaper than a Omniwheel Component , which are mostly easy to change?" :D But James, i really like your Content. Actually i work on my own System for a R2 Droid ^^

[@aerospecies]

hahaha my project lol. I will just copy it then. no need for hard work now. I love this great British mind. Good one James.

[@krbuck]

Reminds me of some warehouse robots I have seen. Would be intuit add spatial awareness and have several collaborate on some tasks.

[@electronic7979]

Excellent

[@SzymekCRX]

8:01 You can assign value1 = -value2 ommiting multiplication for better code performance :)

[@nkronert]

I guess that depends on the hardware implementation of the CPU (multiplication may take only 1 cycle) and the efficiency of the compiler.

[@SzymekCRX]

@@nkronert Yes, perfectly true, but in this case Arduino can make like 1ns difference :) :)

[@soulextinguisher]

It would be neat if it also had the functionality to pivot around each wheel

[@dipakkhatawkar4230]

Hi, Its really impressive. Can you please give the details on how to build the universal remote controller used in this project. Also the detailed connection diagram.

[@ianknowles]

I've needed a robot platform base for some time and this looks like it will fit the bill what's the max load on those motor's? In some ways this is the most useful project you've done.

[@Darkrut]

I wonder if you could eliminate the need for the steering servos by offsetting the wheels from the pivot point at the top and driving the wheel to turn it around the axis.

[@ArtamisBot]

I definitely want to give this one a go 💚

[@rhammond4656]

Hi, really impressed with your channel, I have recently become disabled and my mobility has suffered. I wanted to get into 3d printing and very basic robotics. Would you be able to suggest a reasonably priced radio controller for a novice. Hope you don't mind me asking. Keep up the great work!!

[@gustarrezende]

So cool! Love your little creature! haha

[@adamwolfy4826]

Digital ackerman and digital differentials! Awesome!

[@literate-aside]

This is so impressive.

[@eyadelghareeb898]

would be great as camera robot with this type of movement

[@wakjagner]

I wonder if a version of this, with a lower center of mass, would be the ideal platform for the Really Useful Robot; as it could then slide in closer to objects to manipulate them.

[@Wolficefang]

I’m expecting James to eventually invent a car bot compatible with all his wheels… but maybe part of the fun is designing a new frame every time?

[@Leo99929]

That is very cool!

[@FrancoGrimoldi]

Stunning!

[@DIYwithBatteries]

Wow that's awesome Bro 😃👍

[@shawnshekari]

You sped up the outside wheels but it looks like it is accelerating when you turn, perhaps use your scale factor to slow the inside wheels and keep the outside at 1.0?

[@sasquatchsequoia9455]

Awesome!!! Thanks for sharing!!!

[@StevenIngram]

This might be a fun platform to use to experiment with neural net controls. Let it learn to drive itself and see what it comes up with. :D

[@FilmFactry]

This was excellent!

[@tiagotiagot]

How about calculating the angle and speed of the wheels, by first moving the whole robot (in math) to a new position and orientation based on stick inputs (a single 3 axis stick should do, pushing for translation, and twisting for rotation), then calculating the direction and distance between the current and the new position of each wheel, and using that to decide how to turn the wheels and how fast they need to spin?

[@sorenfox]

Oooh, now do field-oriented control like in FRC!

[@jakelorkin2661]

Joseph?

[@sorenfox]

@@jakelorkin2661 LOL hi Jake