I don't know if you know this, but herbie was ment to me in fan4stic, and was ment to be just like the robot at the start of this video, but with a soccer or basketball!
This really grinds my gears.....need to angle the teeth properly for engagement. Deeper grooves. Gears need to engage properly to get any work out of them. I wish I lived in a higher IQ dimension.
here as well as with some of your gearbox videos (harmonic drive, and I think one of the cycloidal drives), you use flexible filament for grip or ductility. It does do both of those things but it also saps efficiency and puts more load on the motor. Think biking on a flat, soft tire vs an inflated and much harder one. It's a balancing act to get grip and efficiency, but in my experience most 3d printing flexible filaments are too soft for transmission rollers. Maybe try PLA rollers with just a thin rubber tire on the small wheels for grip.
To be fair, compared to the traditional machining, hobby-level printers may be working with tolerances of 0.1 - 0.2mm even when tuned quite well, which is a lot more slop than you'd expect from machined parts. That could have an impact on how well harder rollers grip.
@@syedsulaiman8380 I've got a roll of TPU and have seen lots of other people's prints as well. It may feel "firm" to the touch but its still very soft as compared to the rubber on a tire or drive belt.
@@EgorKaskader True, but even then I think you'd be better off with two hard rollers, with one having a thin TPU tire for more traction and to soak up the loose tolerances. Or at least one soft wheel and one hard wheel to minimize the energy lost to soft, squishy rollers.
What you have here is essentially 2 mecanum wheels engaging intermediate rollers which engage the floor. In my opinion, 2 side-by-side mecanum wheels directly engaging the floor would be better. The only thing you would have to assure is division of the load between the 2 wheels. That can be done with a short equalization lever.
yes, but then the sideways movement would be very reliant on floor traction due to the mencanum wheels. his design would be more relaible on a vareity of surfaces.
An interesting observation, though for self-balancing (his ultimate goal with this) co-axial dual-mecanum wheels are notoriously difficult to use. They also have the down-side of blocking rotation about the axis perpendicular to the line formed between the two wheels, meaning that it wouldn't be able to balance on uneven surfaces (or go up a hill sideways) nearly as well.
@@isaaczoesch158 ANY balancing scheme with 1 or 2 wheels MUST have friction with the surface just to balance. That's why bikers suddenly fall when encountering icy or wet slick roads. The 2 wheels could be guaranteed equal force-bearing by using a lever equalization link (like used in steam locomotives). The largest objection to the 2 side-by-side scheme is the torque about the vertical axis generated when driving the wheels in opposite directions. This could be mitigated by using 3 wheels side-by-side, the outer 2 driven together, & the center one separately. (He might object to this as the vehicle is no longer technically "1 wheel". But the best way to do this is to have real legitimate gearing to drive the smaller wheels. This will be intricate with bevel gears, but doable with his extensive 3D printing equipment & experience if he makes the wheel larger. This would permit the small wheels to be football shaped, giving a true circular outside profile.
I think you may have mis-interpreted part of the patent. Those "teeth" should be helical. All you are doing is reducing the contact area between the driven and drive wheels in your design. Either make them with matching helices, or make them smooth (smooth is probably better unless you understand gears and can machine them more accurately than a 3d print).
you could get more grip between the sub wheels of the omniwheel and the rollers on the side plates by making the subwheels with a 45 degree diamond pattern instead of straight ridges. This way the ridges of the rollers will match up and apply a pushing/pulling force instead of mostly friction
I don't like the patented idea - too much rubbing/friction. The OmBURo concept is much more realistic albeit with room for optimizations. Having the little wheels rolling obliquely against one-another sounds very inefficient.
I think multiple 90° helical gear connections and a large circumference long flexible shaft would produce more friction/be harder to drive. Especially if you have a larger weight on top like a human
Maybe those small driving wheels on sides can have some kind of spiral-shaped gear teeth so they would lock with the bigger wheels. Or it's named helix, not spiral. The point is that the teeth on the smallest wheels can be not just parallel to their axis but go under 45º angle around
One seemingly important difference between your omni-wheel and its inspirations is that the smaller wheels aren't interconnected. I think you are losing a bit of torque and efficiency potential by essentially driving only 1-2 smaller wheels at a time (the ones in contact with ground) rather than all of them.
After the ornithopter videos (no difference in AoA between up/downstroke) and that CMG project (no way to desaturate the gyros), it seems to be a bit of a common theme here that important details of the technology are routinely neglected. Amazing projects nonetheless, even if the results end up hampered by it
but they also seem to use a flexi shaft in both directions- I think they are only strong in one direction and you can damage them easily if you use in the other direction
What if you use interlocking conical gears on the ends of the rim wheel axles. So you get a form of 'bent gear coupler'. These conical gears cab be made for other angles than just 90 degrees like in a differential.
Love it. Do you have any issues with the "strafe" speed being so much slower than the "roll" speed? Seems like it might cause a bit of a control problem
Probably just a case of using just some spare motors, he seems to have just been doing proof of concept here. A faster more powerful motor would do the trick, I think he’s still gotta work out the play in the rollers themselves to work out precision and accuracy with the movement before being able to balance I suspect
if you had a singular ball in the center with the rollers acting as stabilizers instead of the locomotion; could you then have an omni-wheel made from a ball?
Problem with a ball is the dirt you pick Up from the ground getting stuck in the drive, if this wasnt the case Every car in the world would have had globular Tires by now
I wonder if you could print a helical tooth pattern on the outer rollers that the smaller drive rollers can bite into? I assume it would have to be two counter-twisting helices, one for each side of rollers, so effectively the outer rollers would just be knobby.
This will probably not work because the outer wheels are flat/cylindrical and not round/parts of a torus as shown in the patent. But its worth a try, i guess 🤔
I was thinking this as well but having trouble thinking of a profile that would allow engagement with both sets of smaller rollers as well as the ground... Somehow your comment reminded me of a video I saw recently (past three months) with some very interesting knurled or cross-hatched gear profiles that might work.
Idea: mount the two driving wheels at a slight angle relative to the vertical central plane - this way you only get traction - and, more importantly, friction - where you need it: at the bottom
Was about to comment on that as well! But slanted wheels would cause problems with contact patch between the main and driving wheels. The same goes for smaller and eccentrically mounted (towards the bottom) side wheels. Maybe some clever geometry could prevent this?
*While I do like the experiments, I would also like to know what could be useful for?* Because to be honest, sometimes I have no clue of any practical applications In this case, I do know Omni-Wheel applications, but for 1 wheel? Not really.
@@AnnoDominiAD But what's the point of an omni directional car or bike? And if it's so great, why choose the most complicated way to do it? What are we even talking about, here?
@@AnnoDominiAD its too complicadted to be durable and cheap enough and on top of that you have grip problems due to small wheels etc. there are way simpler designs
Less Parts, less complexity, Add it to existing things. Automated shopping cart grabber. Can't get big trains of carts, but a few of them could navigate a car park and use a raising and lowering "arm" to grab carts and move them back into the store. Obviously would need sensors and cameras to avoid cars and people. Patient Transport in Hospitals. Instead of needing an orderly to push a patient around, a "come along" could be used by nurses and doctors to push a patient behind/ahead of them, or could be used to move the mobility challenged around larger facilities. Cheaper to use a base "platform" that is a regular wheelchair, then just have a few bots to do the movement when the patient/practitioner cannot Assisted Drive in various carts. Could counteract turbulence in Airplane carts with a single wheel (saving weight in the process). Could be retrofitted to attach to medical carts and crash cash to aid in reducing energy needed to push and move them, and allow the carts to be physically heavier without affecting the person who needs to push it. Warehouse carts/shelves could have the same idea where one wheel is all thats needed to assist the person moving them aside to get to other items. Mobility scooters. 3 wheel trike style mobility scooters require the operator to lean forward and twist to turn, which is not possible for all users. Joystick controls are easier, but more expensive, requiring 2 motor driven wheels, and they drive different from a typical system. This could combine the driving pattern and simplicity of 3 wheel, with the control scheme and ease of use of joystick drive. While yes, 2 wheels, or more wheels, can be used in all above applications, if this design could use a singular drive motor that's clutched, it can reduce electrical needs and lead to better battery life on all of these items. As well, the space savings can allow this design to be used in weight and space critical operations like on boats, planes, small warehouses, and trains. The cost savings, when a design is industrialized and produced in mass enough, of having 1 omni-wheel versus 2 or 4 could also prove to be superior for multi-unit applications such as the cart grabber or shelf mover in a warehouse. The 5th wheel applications are very diverse for retrofitting, and worth exploring by themselves.
I just found this channel so I don't know anything about past projects but at 1:07 it's mentioned that it turns opposite way it leans. Theirs a video called "Most people don't know how bikes work" by Veritasium It mentions this weird effect is also on bikes. With a slow mo from the front. It could be useful to know these types of things as well as maybe getting slow mo of a unicycle turning.
Is there a way to have a motor fixed to the body that drives only the current bottom wheel? that way there is much less friction and you are not wasting power turning all those other wheels that are not touching the ground. Edit: You would also be able to turn at the same time as you are going forward. It doesn't look to me like the current build does that. Edit 2: Maybe the 3 current bottom wheels so you can turn when going up hill or on uneven terrain.
I thought the same thing. Why drive all the other rollers when the 3 on the bottom are doing all the work. You could just have a friction drive on the bottom 3 rollers.
Nice! 🙂 You should make the teeth of the TPU wheels disgonal so that they mesh like a gear. And you should perhaps make the large rollers more like a barrel, i.e. make the outer diameters somewhat smaller than the diameter in the middle.
I'm always amazed seeing the solutions that the human brain can come up with. I'm also impressed that people are capable of turning such ideas into actual 3D models so they can be manufactured. I am a bit worried about wear and tear on this particular omni wheel.
Use pulleys on the secondary gears to drive the axles of small treads for better ground contact and power delivery, since you could make a high precision set of gears to drive the whole mechanism and keep it protected within the assembly. A tire or pair of tires might work well too…
The complexity is just crazy. To me it feels like a lot of excess mass and bulk when only the tiny wheels at the bottom are actually utilized at any given time. If I were trying to make a single wheel balancing robot, I'd go with a tilted wok shaped wheel. One motor drives the slightly bent axle, while the other motor drives the wheel. The key is to have a small amount of bend, so it's possible to quickly switch direction. It's possible for it to be passively stable, actually, but with a small hemisphere or ball shape it would require active stability.
You seem to know what you're talking about. What is the point of all this? What is the advantage of balancing a robot on a single wheel? Isn't four simple wheels far more cost effective and practical than some Rube Goldberg single wheel contraption? I just don't get the point.
Could you make the small wheels bevel gears, and the larger wheels cross-threading bevel gears (similar to a bolt that can take both left and right hand threaded nuts) to reduce slippage? The math would be a pain though
For in-line two wheel vehicles you actually do have to turn left before you can turn right. This is so counter intuitive most people will fail to grasp it until presented with a physical experiment. For an omniwheel bike this would require you to slip to the left before executing the right turn. You would then have to reverse the procedure to come out of the turn. The required angle of slip can be calculated by balancing the centripetal force for a given radius of turn with the downward force of gravity such that the resulting vector is through the centre of mass to the wheels.
James, for a flexible shaft, you can use a spring, like what you find in a drain snake. They are super tough and can take a lot of torsional load while bent in a radius.
I have seen a RC toy called Air Hogs Upriser that has driven omniwheel and able to self balance on single wheel. Its internal mechanism is more similar to traditional differential drive gears. Not sure whether it is still suitable for more heavy-duty robots.
Love it, love it, love it... sometime you just stumble upon videos that ust gets all the gears in your head turning.... this is def one of them! Thank you so much for psting! Cheers from Dominican Republic.
I'm planning on using a number of these techniques on an upcoming project, to make a replica of the killer tire from the film Rubber. But all the parts have to basically float inside the walls of the tire so that the center remains completely empty. I haven't completely worked out if I want it to be able to lean left and right by shifting gyros or if I can turn the battery box into a servo actuated counterweight in the top and have that be sufficient.
One advantage with the Honda patent version is that all the small wheels are contributing to the rotation force applied the two or three wheels actually supporting the unit. One possible way to make "rope" axle would be to use multiple layers of carbon fibre or kevlar with each layer twisted in opposite directions. You would, of course, have to bind the rope axle . Could possibly use resin.
That "tilting changing the way the wheel steers" is exactly why I decided to dismantle my one wheeled Vespa. I saw the 1000w one that Bel Y Bel made a couple years ago, and decided to try making one with a much stronger 3000w motor, and it was just too hard to control at that speed. You could definitely upscale this Omni wheel to a serious vehicle! I think would only cost about 2000£. Just mount a 42v 1000-2000w brushless ebike motor on each side.
Wow this is an awesome project. Very detailed and time consuming but an amazing project. Cannot imagine how long the prints took where I can only stand to print at max 4 hours at a time lol
Three suggestions. 1) Run the side driver wheel plates on lower centers so only the driven wheels that are in contact with the ground are controlled. This should considerably reduce friction. 2) Barel the outer driven wheels to conform to the other diameter for a smooth ride and concave the driver wheels to mesh. 3) The" teeth" on the driver and driven wheels should be helical.
I invented something similar back in the 90s. It's purpose was to create a robot for robot wars that addressed the number one problem, the control and feedback system. I called it Drive By Wire and you can see something similar in your one wheel robot and stabilised gun mounts in Main Battle Tanks. Yours looks better though, can I use it?
I bet if you used a narrow hub for the drive rollers with a thick you tread you will get the best of both worlds. Flex of the wheel and less play against the axle. Freaking amazing build. I look forward to seeing more.
Ideally you wouldnt need any rotating small wheel apart from the one that touches the ground. So I am thinking either dont drive any wheel but the bottom one at a time or maybe a fixed small wheel that always stays on the contact area and comes into effect by a levering mechanism or similar when there is sideways motion needed?
Hello, thank you for your awesome videos. But I have one question on this particular video. Wouldn't it work better if you had printed diagonal lines on the little hubrolls instead of the straight lines? So that they could line up better with the bigger outputwheels wheels.
I think using worm gear can directly change the spinning direction from the large wheel to the small wheel. Moreover, springs or some elastic plastic can detach the gear of small wheel from the large wheel. When the small wheel press against the ground, the gear bind together again. This may reduce the energy loss due to the friction when moving the useless small wheel
I like the idea, probably you could make it lighter by only keeping traction in the lowest part of the wheel, as the rest of the wheel doesn't really work
What about something that functions more like a unicycle? Have a single drive wheel and allow it to pivot/turn rapidly to pursue a target direction toward underneath the agent's mass. Yes, the mass would rotate in the opposite direction a bit but you could also separate the 'head' from the 'body' and let the body/wheel act on eachother while the head just sorta does its own thing for navigating with sensors/cameras. To roll in a direction it would effectively start rolling the opposite direction to cause itself to begin falling in the direction it wishes to go and then quickly accelerate in that direction to hold itself up. I think that would be a really fun and rewarding challenge for a single-wheeled bot but I'm more of a math/algorithms nerd than an electrical engineering one, and coding up maths is my forte.
I think that would be fun to see, especially considering the more or less torque needed to turn, based on load weight and the surface on which the when is operating. Imagine a 12" wheel trying to quickly turn left/right 45 degrees in grass, or a rubber when with a lot of weight doing it on asphalt.
Question: To get an angled path of travel(forward/backward and left/right simultaneously) would you rotate the two drive gears in opposite directions at differing rates? As in, the difference between the speed of the two rotating gears would be translated into forward/backward motion?
Dude, seriously. How come you are not working for space X, tesla, boston dynamics or related. This is the kind of smart people who has to drive our future!! Good work man!!
What if you made the perimeter wheels spur gears (probably best with truncated teeth) and engaged them with a large wheel, basically a very thin slice of a large diameter worm wheel. Driving the whole assembly would move back and forth, and driving the worm gear backwards or forwards would rotate the rollers, moving side to side. Obviously not great for a dirty environment, but could be an interesting concept to play with.
If the center wheel and the two outer wheels had different centers of rotation then you could adjust it such that only the bottom would be "active" ... thus reducing friction. Great job at conceptualizing this.
I always wondered, how do you efficiently dissipate the heat producedby the motors and processing unit? Are these viabale only for a short term or can they operate for longer periods?
So the drive wheels are held at an angle to the output wheels but the teeth do not need to mesh? Wouldn't that end up in slippage and a loss of energy in the transfer? Or does this have some other advantage?
For this current demo, would a design like a screw that goes both ways work for your central wheel? It would be less binding and always have a point 'flat' against the gear as it pushed on it I'd think. Would be less binding and allow you to tighten up the design a bit ya?
The industrial maintenance training in me see these and thinks, "That's a cool mechanism, but the stuff that will get into it over time is going to kill it." Maybe taking a look at the system that is used for omnidirectional forklifts and then apply a variant of that to this same idea? Like by using two narrow wheels of the same design as the forklift design?
You could probably 'save' some torque if you have the non-contacting wheels disengage from the rollers, allowing for the torque to be placed only the wheels which have contact with a surface. Might be over-engineering it though, as I imagine it wouldn't make a huge difference but it's still a neat idea. I think you could probably do something like this: . If the roller hubs where slightly smaller they wouldn't be able to engage any of the wheels without pressure. . as for how the pressure would be translated, without having to change the lateral wheels; you could design the main hub to allow for small vertical movement of the lateral wheels, so when pressure is applied--either from gravity or weight on contact--the wheels can engage with the roller hubs. . finally you could add fairly low pressure springs to the bars of the lateral wheels within the new cut outs in the main hub, or where ever the vertical movement of the wheels is inserted, allowing for the wheels to 'reset' their position when contact ends - also stopping gravity from engaging the top wheels. Alternatively if you want the lateral wheels to *only* work when contacting the floor--instead of whenever it contacts with *anything*--you could just apply a very small tilt to the roller base so only the bottom few rollers will ever engage.
I think a big solution would be to connect all the shafts of each big grippy gear on the center hub, together with a flexible shaft like in the other design. This way, when two gears do not interface properly, this effect is mitigated by all the other gears that do make full contact. This of course also asks for bearings next to each gear in the central hub, so this might become a heavy wheel by itself. But I think it's worth upscaling it a bit and going all the way.
1:09 It's not an unexpected twist. Look into how motorcycles steer. The wheel falls to the outside of the direction turned. The same thing happens in a car, except you have another set of wheels to catch the fall. This is why something hanging from a rear view mirror goes to the outside of a turn.
Can you print the small drive wheels with a worm gear like profile interlocking the outside wheels? And maybe make the outside wheels a bit curved, ball like, and the small drive wheels concave worm gears? (Or would that hinder if the drive wheels hop over to the next outer wheel?)
You could also tilt the axis of the driving hubs so that the hub wheels only contact the main wheels at the bottom, since that's the only place it matters. This would likely significantly reduce the friction losses in the system.
I seen a video on youtube bout a bolt with two way threads.. That is to say threads, which would allow you to turn a nut clockwise or counter clockwise and still tighten it or loosen it. instead of looking more like a screw. if you used this same idea on you main drive wheels, you coold put bi-directional "teeth/threads" into the main wheels that would fit the rollers on each side better and maximize contact and grip between the rollers and the main hub.
Hydraulic tubing makes a good flexible shaft that is cost effective. I've successfully used it as a 2 foot diameter flex shaft. I assume smaller diameter tubing would work in this case.
I cannot believe I live in a time where this kind of bizarrely niche and amazing information is freely available to anyone..... There is hope for humanity.
Fabulous work, congratulations. I have a feeling I'm looking at a summary of years of work. It's a wonderful new world we live in with 3D printers, cheap electronics and control systems.
There's a simpler way to do that. You have a large wheel that's open on the inside as to house a L-R driving mechanism. The hub would be biased to the opposite side (like a car wheel) or use a hub less wheel with the rollers around it. The bottom where ground contact is made sits just 1 or 2 omni wheels that drives only the ground contact rollers left or right. *The reason it's an omni wheel is so the forward/reverse rotation of the large wheel doesn't cause friction for the smaller wheel which drives the left/right motion. *You may want 2 so that thetrnsition between rollers is seamless.
What if you have the angled wheels shaped with inverse curvature, like this )|( where the | is the axle; would that allow for a more reliable transmission of motion ?
do you think it would work better if you made the outside smaller wheels traction at a 45 degree angle also so it lined up better with the traction of the bigger center wheels?
Is there a way to do this without friction? Some sort of way to isolate the drive mechanism from touching the road. Otherwise, you'd get grit and debris from the road stuck in the mechanism.
Hey instead of those small outer wheels to turn the other bigger wheel, do you think gears In place of small wheels to help turn bigger wheels could help? I like the design btw
the problem this will never work in real cars, because it has to many parts, and it would be to expensive and difficult to replace every part, what makes it not practical. But it is a nice concept.
Nice experiment. Did you think of the side rollers to change from 0-degrees knurl to 45-degrees (ie helical) knurl? This enable all knurls mesh with each other and does not need to rely on friction. What do you think?
The terrifying speed of digrodation. Two hundred years ago , two-wheeled cars were made, railway cars on one rail. Not on the rail that occupies 2.3 of the width of the train itself, but a simple rail. And all this on a gyroscope. But then everyone forgot about it, and the technology went into stabilizers for ship guns.
Very cool! I wonder if there would be any benefits from making the drive hubs smaller, and offsetting them so the little gears only engage with the larger ones that are in contact with the floor. Maybe the little hubs could also be more perpendicular to the main hub (more parallel with the floor) so they could squeeze the driving gears harder, while also reducing the lateral twisting on the little hubs.
9:59 - it is strange that he didn't make those smaller gears so they would have the "furrows" in an angle, which, I suppose, will increase the grip with the bigger gears. Am I missing something in here?
You can find full core rubber wheels for bicycles that will do exactly what you need them to do. The single piece rubber is the simpler/smarter solution.
Check out part 2 - will it balance on one wheel? ua-cam.com/video/B_bXW2vfkm0/v-deo.html
I don't know if you know this, but herbie was ment to me in fan4stic, and was ment to be just like the robot at the start of this video, but with a soccer or basketball!
ua-cam.com/video/-jqBKZkPn80/v-deo.html
This really grinds my gears.....need to angle the teeth properly for engagement. Deeper grooves. Gears need to engage properly to get any work out of them. I wish I lived in a higher IQ dimension.
I really couldn't figure out the potential advantages of a single complicated wheel over 4 simple wheels... What is this going to be applied to?
The Mel science web site is garbage. Nice Clear Pricing!!!! No request for subsription or info... WTF
The Honda U3-X looks like a chair an evil mastermind would have
I want several of it and its further evolved descendants.
It looks like something a hilarious Bond villain caricature played by Mike Myers or David Mitchell would have.
Exactly what I was thinking
OK Go did an entire music video dancing with these. XD
We miss the option for the evil-sidekick-cat 😂
here as well as with some of your gearbox videos (harmonic drive, and I think one of the cycloidal drives), you use flexible filament for grip or ductility. It does do both of those things but it also saps efficiency and puts more load on the motor. Think biking on a flat, soft tire vs an inflated and much harder one. It's a balancing act to get grip and efficiency, but in my experience most 3d printing flexible filaments are too soft for transmission rollers. Maybe try PLA rollers with just a thin rubber tire on the small wheels for grip.
To be fair, compared to the traditional machining, hobby-level printers may be working with tolerances of 0.1 - 0.2mm even when tuned quite well, which is a lot more slop than you'd expect from machined parts. That could have an impact on how well harder rollers grip.
Have you ever used TPU depending on the infill u can't make it very hard
I don't think ur they are as soft as you think they are
@@syedsulaiman8380 I've got a roll of TPU and have seen lots of other people's prints as well. It may feel "firm" to the touch but its still very soft as compared to the rubber on a tire or drive belt.
@@EgorKaskader True, but even then I think you'd be better off with two hard rollers, with one having a thin TPU tire for more traction and to soak up the loose tolerances. Or at least one soft wheel and one hard wheel to minimize the energy lost to soft, squishy rollers.
Listen, you don’t know a thing
What you have here is essentially 2 mecanum wheels engaging intermediate rollers which engage the floor. In my opinion, 2 side-by-side mecanum wheels directly engaging the floor would be better. The only thing you would have to assure is division of the load between the 2 wheels. That can be done with a short equalization lever.
Yes but then there would be an extra torque when the two side-by-side mecanum wheels are rotating in the opposite direction
yes, but then the sideways movement would be very reliant on floor traction due to the mencanum wheels. his design would be more relaible on a vareity of surfaces.
An interesting observation, though for self-balancing (his ultimate goal with this) co-axial dual-mecanum wheels are notoriously difficult to use. They also have the down-side of blocking rotation about the axis perpendicular to the line formed between the two wheels, meaning that it wouldn't be able to balance on uneven surfaces (or go up a hill sideways) nearly as well.
here's a crazy idea,. have two mecanum wheels, but put one inside the other!
@@isaaczoesch158 ANY balancing scheme with 1 or 2 wheels MUST have friction with the surface just to balance. That's why bikers suddenly fall when encountering icy or wet slick roads. The 2 wheels could be guaranteed equal force-bearing by using a lever equalization link (like used in steam locomotives). The largest objection to the 2 side-by-side scheme is the torque about the vertical axis generated when driving the wheels in opposite directions. This could be mitigated by using 3 wheels side-by-side, the outer 2 driven together, & the center one separately. (He might object to this as the vehicle is no longer technically "1 wheel".
But the best way to do this is to have real legitimate gearing to drive the smaller wheels. This will be intricate with bevel gears, but doable with his extensive 3D printing equipment & experience if he makes the wheel larger. This would permit the small wheels to be football shaped, giving a true circular outside profile.
I think you may have mis-interpreted part of the patent.
Those "teeth" should be helical.
All you are doing is reducing the contact area between the driven and drive wheels in your design.
Either make them with matching helices, or make them smooth (smooth is probably better unless you understand gears and can machine them more accurately than a 3d print).
There's many things odd about this design.
you could get more grip between the sub wheels of the omniwheel and the rollers on the side plates by making the subwheels with a 45 degree diamond pattern instead of straight ridges. This way the ridges of the rollers will match up and apply a pushing/pulling force instead of mostly friction
@Get on the cross and don’t look back Stop spamming
@@justsomeguywithoutamustach7151 the original comment may be deleted, but based on the name I have a pretty good idea of what was posted
I don't like the patented idea - too much rubbing/friction. The OmBURo concept is much more realistic albeit with room for optimizations. Having the little wheels rolling obliquely against one-another sounds very inefficient.
...so does the flexible shaft and the 90° gears.
would to good to be true if this is every day life some day.
@@doopfdeckel
"would to good to be true if this is every day life some day"
Lol wut?
I think multiple 90° helical gear connections and a large circumference long flexible shaft would produce more friction/be harder to drive. Especially if you have a larger weight on top like a human
@@生マグロ-s8e or your mom...
hehe sorry i couldn't help myself
@@生マグロ-s8e they both use flex shafts in the original designs he showed.
Your 'prototypes' and 'test devices' are amazing accomplishments all on their own! Well done!
Maybe those small driving wheels on sides can have some kind of spiral-shaped gear teeth so they would lock with the bigger wheels. Or it's named helix, not spiral. The point is that the teeth on the smallest wheels can be not just parallel to their axis but go under 45º angle around
One seemingly important difference between your omni-wheel and its inspirations is that the smaller wheels aren't interconnected.
I think you are losing a bit of torque and efficiency potential by essentially driving only 1-2 smaller wheels at a time (the ones in contact with ground) rather than all of them.
After the ornithopter videos (no difference in AoA between up/downstroke) and that CMG project (no way to desaturate the gyros), it seems to be a bit of a common theme here that important details of the technology are routinely neglected. Amazing projects nonetheless, even if the results end up hampered by it
but they also seem to use a flexi shaft in both directions- I think they are only strong in one direction and you can damage them easily if you use in the other direction
What if you use interlocking conical gears on the ends of the rim wheel axles. So you get a form of 'bent gear coupler'.
These conical gears cab be made for other angles than just 90 degrees like in a differential.
Love it. Do you have any issues with the "strafe" speed being so much slower than the "roll" speed? Seems like it might cause a bit of a control problem
@C l a r a ♥️ Silence, thot bot. Men are talking about stuff that actually matters.
Probably just a case of using just some spare motors, he seems to have just been doing proof of concept here. A faster more powerful motor would do the trick, I think he’s still gotta work out the play in the rollers themselves to work out precision and accuracy with the movement before being able to balance I suspect
So, if the two motors engage at the same direction but different speed, does it move diagonally?
Yes
if you had a singular ball in the center with the rollers acting as stabilizers instead of the locomotion; could you then have an omni-wheel made from a ball?
ua-cam.com/video/byUo4EagXu0/v-deo.html that would be this with a smaller ball
Problem with a ball is the dirt you pick Up from the ground getting stuck in the drive, if this wasnt the case Every car in the world would have had globular Tires by now
@@johnpetters3328 Brushes and maybe compressed air
@@jennalove6755 Yeah, seems like it would be pretty simple to push air through from the top to flush out and even cool the wheel if need be.
think we need to show more appreciation to Honda who are just inventive geniuses.
Honda's R&D dept christmas parties must be fuckin' wild.
I wonder if you could print a helical tooth pattern on the outer rollers that the smaller drive rollers can bite into? I assume it would have to be two counter-twisting helices, one for each side of rollers, so effectively the outer rollers would just be knobby.
This will probably not work because the outer wheels are flat/cylindrical and not round/parts of a torus as shown in the patent. But its worth a try, i guess 🤔
I was thinking this as well but having trouble thinking of a profile that would allow engagement with both sets of smaller rollers as well as the ground... Somehow your comment reminded me of a video I saw recently (past three months) with some very interesting knurled or cross-hatched gear profiles that might work.
Idea: mount the two driving wheels at a slight angle relative to the vertical central plane - this way you only get traction - and, more importantly, friction - where you need it: at the bottom
It strikes me that the whole mechanism could be simpler if you were only concerned about driving the bottom 3 or 4 rollers.
Yes, the only reason to drive all the wheels is simplicity of design
Was about to comment on that as well! But slanted wheels would cause problems with contact patch between the main and driving wheels. The same goes for smaller and eccentrically mounted (towards the bottom) side wheels. Maybe some clever geometry could prevent this?
@@huzeff I'm sure the geometry would have to adapt - but that's not an issue for James
You're seriously a genius man... I wish I had your level of consistent intellectual persistence
*While I do like the experiments, I would also like to know what could be useful for?*
Because to be honest, sometimes I have no clue of any practical applications
In this case, I do know Omni-Wheel applications, but for 1 wheel? Not really.
2 omni wheels = omni directional bike
4 omni wheels = omni directional cars
etc
That's what I was wondering, the whole way through. I couldn't think of anything to do with it.
@@AnnoDominiAD But what's the point of an omni directional car or bike? And if it's so great, why choose the most complicated way to do it? What are we even talking about, here?
@@AnnoDominiAD its too complicadted to be durable and cheap enough
and on top of that you have grip problems due to small wheels
etc.
there are way simpler designs
Less Parts, less complexity, Add it to existing things.
Automated shopping cart grabber. Can't get big trains of carts, but a few of them could navigate a car park and use a raising and lowering "arm" to grab carts and move them back into the store. Obviously would need sensors and cameras to avoid cars and people.
Patient Transport in Hospitals. Instead of needing an orderly to push a patient around, a "come along" could be used by nurses and doctors to push a patient behind/ahead of them, or could be used to move the mobility challenged around larger facilities. Cheaper to use a base "platform" that is a regular wheelchair, then just have a few bots to do the movement when the patient/practitioner cannot
Assisted Drive in various carts. Could counteract turbulence in Airplane carts with a single wheel (saving weight in the process). Could be retrofitted to attach to medical carts and crash cash to aid in reducing energy needed to push and move them, and allow the carts to be physically heavier without affecting the person who needs to push it. Warehouse carts/shelves could have the same idea where one wheel is all thats needed to assist the person moving them aside to get to other items.
Mobility scooters. 3 wheel trike style mobility scooters require the operator to lean forward and twist to turn, which is not possible for all users. Joystick controls are easier, but more expensive, requiring 2 motor driven wheels, and they drive different from a typical system. This could combine the driving pattern and simplicity of 3 wheel, with the control scheme and ease of use of joystick drive.
While yes, 2 wheels, or more wheels, can be used in all above applications, if this design could use a singular drive motor that's clutched, it can reduce electrical needs and lead to better battery life on all of these items. As well, the space savings can allow this design to be used in weight and space critical operations like on boats, planes, small warehouses, and trains. The cost savings, when a design is industrialized and produced in mass enough, of having 1 omni-wheel versus 2 or 4 could also prove to be superior for multi-unit applications such as the cart grabber or shelf mover in a warehouse. The 5th wheel applications are very diverse for retrofitting, and worth exploring by themselves.
I just found this channel so I don't know anything about past projects but at 1:07 it's mentioned that it turns opposite way it leans.
Theirs a video called "Most people don't know how bikes work" by Veritasium
It mentions this weird effect is also on bikes. With a slow mo from the front. It could be useful to know these types of things as well as maybe getting slow mo of a unicycle turning.
Is there a way to have a motor fixed to the body that drives only the current bottom wheel? that way there is much less friction and you are not wasting power turning all those other wheels that are not touching the ground.
Edit: You would also be able to turn at the same time as you are going forward. It doesn't look to me like the current build does that.
Edit 2: Maybe the 3 current bottom wheels so you can turn when going up hill or on uneven terrain.
I thought the same thing. Why drive all the other rollers when the 3 on the bottom are doing all the work. You could just have a friction drive on the bottom 3 rollers.
It never ceases to amaze me how many sponsors you can get in a video .
Nice! 🙂
You should make the teeth of the TPU wheels disgonal so that they mesh like a gear. And you should perhaps make the large rollers more like a barrel, i.e. make the outer diameters somewhat smaller than the diameter in the middle.
Amazing how old the gyroscope is and never let us down and used so proficiently today in many different fields
I wonder if there is a special gear tooth geometry that would increase the contact between the red rollers.
This is so cool, that guy just floating about on a chair in all directions is awesome I can't wait to see the next iteration...cheers.
I'm always amazed seeing the solutions that the human brain can come up with.
I'm also impressed that people are capable of turning such ideas into actual 3D models so they can be manufactured.
I am a bit worried about wear and tear on this particular omni wheel.
Just wait until you see the solutions that non-human brains come up with....
@@Paul-rs4gd an air hockey table? 😉
Use pulleys on the secondary gears to drive the axles of small treads for better ground contact and power delivery, since you could make a high precision set of gears to drive the whole mechanism and keep it protected within the assembly. A tire or pair of tires might work well too…
This guy has more sponsors than the super bowl.
Does it need to drive all the lateral wheels?only the bottom 3 or 4 need power. Right?
Brilliant concept, great execution as usual 👍
The complexity is just crazy. To me it feels like a lot of excess mass and bulk when only the tiny wheels at the bottom are actually utilized at any given time.
If I were trying to make a single wheel balancing robot, I'd go with a tilted wok shaped wheel. One motor drives the slightly bent axle, while the other motor drives the wheel. The key is to have a small amount of bend, so it's possible to quickly switch direction.
It's possible for it to be passively stable, actually, but with a small hemisphere or ball shape it would require active stability.
You seem to know what you're talking about. What is the point of all this? What is the advantage of balancing a robot on a single wheel? Isn't four simple wheels far more cost effective and practical than some Rube Goldberg single wheel contraption? I just don't get the point.
@@hxhdfjifzirstc894 It's just fun to try and engineer. There isn't really a practical point.
You always have great socks, James. Completely off-topic, but I just had to point it out.
Could you make the small wheels bevel gears, and the larger wheels cross-threading bevel gears (similar to a bolt that can take both left and right hand threaded nuts) to reduce slippage? The math would be a pain though
For in-line two wheel vehicles you actually do have to turn left before you can turn right. This is so counter intuitive most people will fail to grasp it until presented with a physical experiment.
For an omniwheel bike this would require you to slip to the left before executing the right turn. You would then have to reverse the procedure to come out of the turn. The required angle of slip can be calculated by balancing the centripetal force for a given radius of turn with the downward force of gravity such that the resulting vector is through the centre of mass to the wheels.
James, for a flexible shaft, you can use a spring, like what you find in a drain snake. They are super tough and can take a lot of torsional load while bent in a radius.
* Skips to 8:09 *
My Smooth Brain: Haha Cokie Cola Lids
I have seen a RC toy called Air Hogs Upriser that has driven omniwheel and able to self balance on single wheel. Its internal mechanism is more similar to traditional differential drive gears. Not sure whether it is still suitable for more heavy-duty robots.
can i just say, has antone noticed that this guy has the most ad sponsors than any video that anyone has watched, well for me that is. anyone agree?
Love it, love it, love it... sometime you just stumble upon videos that ust gets all the gears in your head turning.... this is def one of them!
Thank you so much for psting! Cheers from Dominican Republic.
I'm planning on using a number of these techniques on an upcoming project, to make a replica of the killer tire from the film Rubber. But all the parts have to basically float inside the walls of the tire so that the center remains completely empty. I haven't completely worked out if I want it to be able to lean left and right by shifting gyros or if I can turn the battery box into a servo actuated counterweight in the top and have that be sufficient.
One advantage with the Honda patent version is that all the small wheels are contributing to the rotation force applied the two or three wheels actually supporting the unit.
One possible way to make "rope" axle would be to use multiple layers of carbon fibre or kevlar with each layer twisted in opposite directions.
You would, of course, have to bind the rope axle . Could possibly use resin.
That "tilting changing the way the wheel steers" is exactly why I decided to dismantle my one wheeled Vespa. I saw the 1000w one that Bel Y Bel made a couple years ago, and decided to try making one with a much stronger 3000w motor, and it was just too hard to control at that speed. You could definitely upscale this Omni wheel to a serious vehicle! I think would only cost about 2000£. Just mount a 42v 1000-2000w brushless ebike motor on each side.
Wow this is an awesome project. Very detailed and time consuming but an amazing project. Cannot imagine how long the prints took where I can only stand to print at max 4 hours at a time lol
Three suggestions.
1) Run the side driver wheel plates on lower centers so only the driven wheels that are in contact with the ground are controlled. This should considerably reduce friction.
2) Barel the outer driven wheels to conform to the other diameter for a smooth ride and concave the driver wheels to mesh.
3) The" teeth" on the driver and driven wheels should be helical.
I invented something similar back in the 90s.
It's purpose was to create a robot for robot wars that addressed the number one problem, the control and feedback system.
I called it Drive By Wire and you can see something similar in your one wheel robot and stabilised gun mounts in Main Battle Tanks.
Yours looks better though, can I use it?
Hello Sir. What is the wheel you used for demonstration at 1:58 . Also can you please provide a link ?
I bet if you used a narrow hub for the drive rollers with a thick you tread you will get the best of both worlds. Flex of the wheel and less play against the axle.
Freaking amazing build. I look forward to seeing more.
Ideally you wouldnt need any rotating small wheel apart from the one that touches the ground. So I am thinking either dont drive any wheel but the bottom one at a time or maybe a fixed small wheel that always stays on the contact area and comes into effect by a levering mechanism or similar when there is sideways motion needed?
Hello, thank you for your awesome videos. But I have one question on this particular video.
Wouldn't it work better if you had printed diagonal lines on the little hubrolls instead of the straight lines? So that they could line up better with the bigger outputwheels wheels.
I think using worm gear can directly change the spinning direction from the large wheel to the small wheel. Moreover, springs or some elastic plastic can detach the gear of small wheel from the large wheel. When the small wheel press against the ground, the gear bind together again. This may reduce the energy loss due to the friction when moving the useless small wheel
I like the idea, probably you could make it lighter by only keeping traction in the lowest part of the wheel, as the rest of the wheel doesn't really work
2:14 I'M FUCKING INVINCIBLE
What about something that functions more like a unicycle? Have a single drive wheel and allow it to pivot/turn rapidly to pursue a target direction toward underneath the agent's mass. Yes, the mass would rotate in the opposite direction a bit but you could also separate the 'head' from the 'body' and let the body/wheel act on eachother while the head just sorta does its own thing for navigating with sensors/cameras. To roll in a direction it would effectively start rolling the opposite direction to cause itself to begin falling in the direction it wishes to go and then quickly accelerate in that direction to hold itself up. I think that would be a really fun and rewarding challenge for a single-wheeled bot but I'm more of a math/algorithms nerd than an electrical engineering one, and coding up maths is my forte.
I think that would be fun to see, especially considering the more or less torque needed to turn, based on load weight and the surface on which the when is operating. Imagine a 12" wheel trying to quickly turn left/right 45 degrees in grass, or a rubber when with a lot of weight doing it on asphalt.
That unexpected twisting was covered in that veritasium video with the bike that can (not/only) steer right.
Question: To get an angled path of travel(forward/backward and left/right simultaneously) would you rotate the two drive gears in opposite directions at differing rates? As in, the difference between the speed of the two rotating gears would be translated into forward/backward motion?
Dude, seriously. How come you are not working for space X, tesla, boston dynamics or related. This is the kind of smart people who has to drive our future!! Good work man!!
Could you use small outrunner motors as hub motors for each roller or use 2-4 of them connected to each other via segments of flexible drive cable?
I'm just a simple guy, I discover a channel that blows my mind in the first 10 seconds of the first video and I like and subscribe immediately
What if you made the perimeter wheels spur gears (probably best with truncated teeth) and engaged them with a large wheel, basically a very thin slice of a large diameter worm wheel. Driving the whole assembly would move back and forth, and driving the worm gear backwards or forwards would rotate the rollers, moving side to side. Obviously not great for a dirty environment, but could be an interesting concept to play with.
1:12 "steer in the opposite way it was leaning" - to start a turn on a bicycle, you need to give a brief push to the other handlebar.
If the center wheel and the two outer wheels had different centers of rotation then you could adjust it such that only the bottom would be "active" ... thus reducing friction. Great job at conceptualizing this.
I always wondered, how do you efficiently dissipate the heat producedby the motors and processing unit? Are these viabale only for a short term or can they operate for longer periods?
They run cold in this application
I was watching and thinking "what about a reaction wheel" and then you got to that. These are very clever and cool machines you built!
So the drive wheels are held at an angle to the output wheels but the teeth do not need to mesh? Wouldn't that end up in slippage and a loss of energy in the transfer? Or does this have some other advantage?
For this current demo, would a design like a screw that goes both ways work for your central wheel? It would be less binding and always have a point 'flat' against the gear as it pushed on it I'd think. Would be less binding and allow you to tighten up the design a bit ya?
The industrial maintenance training in me see these and thinks, "That's a cool mechanism, but the stuff that will get into it over time is going to kill it." Maybe taking a look at the system that is used for omnidirectional forklifts and then apply a variant of that to this same idea? Like by using two narrow wheels of the same design as the forklift design?
do you need that nut for the center wheel? Seems like it could hold the wheel via tension.
It just holds the axle square
I suspect it would be a good idea to redesign this so that the contact patch is not the same surface as the drive gearing.
You could probably 'save' some torque if you have the non-contacting wheels disengage from the rollers, allowing for the torque to be placed only the wheels which have contact with a surface. Might be over-engineering it though, as I imagine it wouldn't make a huge difference but it's still a neat idea.
I think you could probably do something like this:
. If the roller hubs where slightly smaller they wouldn't be able to engage any of the wheels without pressure.
. as for how the pressure would be translated, without having to change the lateral wheels; you could design the main hub to allow for small vertical movement of the lateral wheels, so when pressure is applied--either from gravity or weight on contact--the wheels can engage with the roller hubs.
. finally you could add fairly low pressure springs to the bars of the lateral wheels within the new cut outs in the main hub, or where ever the vertical movement of the wheels is inserted, allowing for the wheels to 'reset' their position when contact ends - also stopping gravity from engaging the top wheels.
Alternatively if you want the lateral wheels to *only* work when contacting the floor--instead of whenever it contacts with *anything*--you could just apply a very small tilt to the roller base so only the bottom few rollers will ever engage.
I think a big solution would be to connect all the shafts of each big grippy gear on the center hub, together with a flexible shaft like in the other design. This way, when two gears do not interface properly, this effect is mitigated by all the other gears that do make full contact. This of course also asks for bearings next to each gear in the central hub, so this might become a heavy wheel by itself. But I think it's worth upscaling it a bit and going all the way.
1:09 It's not an unexpected twist. Look into how motorcycles steer. The wheel falls to the outside of the direction turned. The same thing happens in a car, except you have another set of wheels to catch the fall. This is why something hanging from a rear view mirror goes to the outside of a turn.
I ordered the MEL Science kit from the advert. They sent a turd in a jar and a magnifying glass, had hours of fun with it. Yay MEL Science!
Can you print the small drive wheels with a worm gear like profile interlocking the outside wheels?
And maybe make the outside wheels a bit curved, ball like, and the small drive wheels concave worm gears? (Or would that hinder if the drive wheels hop over to the next outer wheel?)
I've gotta show this to my FTC team! They're gonna love this, thanks for the amazing video
You could also tilt the axis of the driving hubs so that the hub wheels only contact the main wheels at the bottom, since that's the only place it matters. This would likely significantly reduce the friction losses in the system.
I seen a video on youtube bout a bolt with two way threads.. That is to say threads, which would allow you to turn a nut clockwise or counter clockwise and still tighten it or loosen it. instead of looking more like a screw. if you used this same idea on you main drive wheels, you coold put bi-directional "teeth/threads" into the main wheels that would fit the rollers on each side better and maximize contact and grip between the rollers and the main hub.
Could you reprint the smaller rollers with a diagonal tread pattern that would then mesh with the lines on the big roller? Kinda like a tiny gear?
Hydraulic tubing makes a good flexible shaft that is cost effective. I've successfully used it as a 2 foot diameter flex shaft. I assume smaller diameter tubing would work in this case.
I cannot believe I live in a time where this kind of bizarrely niche and amazing information is freely available to anyone..... There is hope for humanity.
I've rode U3-X. It's looks magical but feels surprisingly natural to use. Honda totally should mass market it
Fabulous work, congratulations. I have a feeling I'm looking at a summary of years of work. It's a wonderful new world we live in with 3D printers, cheap electronics and control systems.
Which tpu did you use for the wheels at 6:30 ?
How many sponsors can you have per video?
James: "Yes"
There's a simpler way to do that. You have a large wheel that's open on the inside as to house a L-R driving mechanism. The hub would be biased to the opposite side (like a car wheel) or use a hub less wheel with the rollers around it. The bottom where ground contact is made sits just 1 or 2 omni wheels that drives only the ground contact rollers left or right.
*The reason it's an omni wheel is so the forward/reverse rotation of the large wheel doesn't cause friction for the smaller wheel which drives the left/right motion.
*You may want 2 so that thetrnsition between rollers is seamless.
What if you have the angled wheels shaped with inverse curvature, like this )|( where the | is the axle; would that allow for a more reliable transmission of motion ?
do you think it would work better if you made the outside smaller wheels traction at a 45 degree angle also so it lined up better with the traction of the bigger center wheels?
Is there a way to do this without friction? Some sort of way to isolate the drive mechanism from touching the road. Otherwise, you'd get grit and debris from the road stuck in the mechanism.
Hey instead of those small outer wheels to turn the other bigger wheel, do you think gears In place of small wheels to help turn bigger wheels could help? I like the design btw
the problem this will never work in real cars, because it has to many parts, and it would be to expensive and difficult to replace every part, what makes it not practical. But it is a nice concept.
Whats up with the gears 12:00 shouldn't u have made the smaller gears like have angled teeth?
Nice experiment. Did you think of the side rollers to change from 0-degrees knurl to 45-degrees (ie helical) knurl? This enable all knurls mesh with each other and does not need to rely on friction. What do you think?
what about using heat shrink rubber around the wheels to give more grip on floor surface and interacting with other wheels?
Really nice concept.wish to see how BMW AVTR wheels works..james should make one....
This Honda exhibition looks absolutely RIDICULOUS! I'd rather use my legs and WALK or ride a bike.
Hello Mr James, good work. Did you build that joystick? Where I can find it?
The terrifying speed of digrodation. Two hundred years ago
, two-wheeled cars were made, railway cars on one rail. Not on the rail that occupies 2.3 of the width of the train itself, but a simple rail. And all this on a gyroscope. But then everyone forgot about it, and the technology went into stabilizers for ship guns.
Very cool! I wonder if there would be any benefits from making the drive hubs smaller, and offsetting them so the little gears only engage with the larger ones that are in contact with the floor. Maybe the little hubs could also be more perpendicular to the main hub (more parallel with the floor) so they could squeeze the driving gears harder, while also reducing the lateral twisting on the little hubs.
9:59 - it is strange that he didn't make those smaller gears so they would have the "furrows" in an angle, which, I suppose, will increase the grip with the bigger gears. Am I missing something in here?
You can find full core rubber wheels for bicycles that will do exactly what you need them to do. The single piece rubber is the simpler/smarter solution.