Swerve Drives, Omni Wheels, X Drives, and a plea to stop trolling

I was in the process of typing this on the now-locked Swerve Drive thread, and have since been asked to post it here.

Alright, I’ve been summoned to write some sort of end-all-be-all comment so you guys will stop arguing and trolling members of the forum.

First off - Stop being trolls. People are here asking genuine questions, and expect genuine answers. You all are in high school or college, and are mature enough to provide thoughtful, well-thought out responses. There is no point in creating drama out of nothing, especially here, on a robotics forum where people come to learn.


(Above: Not sure who made this, but it was taken from the now-locked Swerve Drive post)

Anyways.
Swerve Drives, while undoubtedly useful if built correctly, are typically considered a non-viable drive train type due to their excessive motor requirement and high complexity. Your attached image is obviously a good example of a swerve drive and the cost-prohibitive use of 8 393 Motors. Keep in mind that when building, you are allowed to use a maximum of 12 393 motors OR 8 V5 motors. Choosing to use 8 393 motors solely on your drive train restricts you to using 4 motors for every other mechanism (without the use of a transmission, which, again, adds unnecessary complexity); whereas building an 8-motor V5 swerve drive does not leave you any leftover motors to power anything else.

After this was posted, I was asked by Kyle | 81818X to mention that swerve drives have the ability to provide the robot’s full torque force in any applicable direction (given that the wheels have 360 degrees of motion)


(Above: Differential Swerve Drive by Kyle from team 81818X)

Obviously the linked CAD model of Kyle’s differential swerve now seems tempting to you. You’re probably thinking you have the gears and the metal and you probably do. It doesn’t look terribly hard to build. But such a drive is A) alot harder to build than it looks and B) is likely to have serious friction problems without any sort of lubrication and C) if both sides are not build perfectly it simply isn’t going to drive straight without a mechanical or programmed jerry-rigged solution. Not exactly ideal.

Might I also note that swerve drives are typically considered because they allow for increased accuracy while turning, as well as variable speed and torque? Omni wheels, when used properly (and they are used by most teams you’ll see at Worlds), will allow any robot to pretty much turn on a dime. And while there are arguments for variable speed and torque, those things can be somewhat achieved using A) programming or B) gearing.


(Above: Not sure who made this)

Which brings me to the final discussed topic, X-Drives, Holonomic Drives, “Bendy Drives”, whatever. Every year there is always a debate about using an x-drive versus a tank drive, and I’m sure a quick search on the forum will yield plenty of information regarding the topic. X drives utilize 4 omni-wheels positioned at a 45* angle to the robot’s side profile to allow the robot to strafe and move at an increased speed than tank drives. (https://aura.org.nz/why-is-x-drive-faster/ for more information on this property.) X-drives can use four motors (and there are many VexU robots that utilized 8-motor x-drives as well), and are quite simple to build compared to a swerve or swerve differential (though tank drive is still the simplest option). X drives may retain slight friction issues due to construction, but when built properly, they do serve as an excellent option for teams looking for more speed and maneuverability.

TL;DR - Stop starting drama and act like adults. You all should be mature enough to help people with their questions.
Swerve drives can grant variable speed and torque, at a massively increased motor cost, robot complexity, and additional programming challenges. Most people would say to look at other drive options.
Omni wheels are a great substitute for the turning capabilities of a swerve drive without any of the complexity. They’re used by most teams you’ll see at state, national, and Worlds competitions.
-X/Holonomic/“Bendy” drives use 4 omni wheels positioned at a 45 angle to allow a robot to strafe and move at a higher speed than a tank drive. They are an excellent option for teams looking for more speed and maneuverability, but there will always be an annual debate about their viability compared to a typical tank drive.

63 Likes

I applaud you for typing this out. I normally just ignore the trolls and occasionally flag posts, but this was much needed.

30 Likes

Umm… there are middle schoolers here too, but I agree with the last part of this comment. Although this:

was a little harsh.

I respect your point of view, but I think it could have been stated differently. By the way, I think this post should be put in one of the wiki topics. It really has a thorough breakdown of swerve drives. I can see that you have put quite a bit of time, thought, and effort into this. Nice job!

4 Likes

Given the situation, I think that being a little harsh is fine. And I’m not really sure why you see that as “harsh”. It’s not that difficult to have a little more self control.

14 Likes

I just read the other thread… I think. It’s the one that DRow closed 2 hours ago right? Yeah. that thread got out of hand. Given THAT thread, you have a fair point. Forget what I said. Nice job on the research!

2 Likes

Yeah I read that thread like 20 minutes and just facepalmed.

4 Likes

Great post @MyrddinEmyrs!

When selecting your drive you always have to consider Speed, Torque (that are inversely related due to the limited power Speed*Torque=Power), and the directions in which they need to be applied the most.

In the ordinary “tank drive” all your power (speed*torque) is concentrated in the forward/backward direction.

When you choose a holonomic drive over the tank drive setup you pay for the ability to instantaneously go in any direction by the added power losses to friction in the rollers and, in case of mecanums, between the wheels and the bearings on the sides.

Since this topic comes up regularly, I would like to quote my post from the last year:

Also, mecanum drives are very similar to the traditional holonomic x-drives in their power vs direction distribution diagram:

12 Likes

Call it by its real name. There is no bendy drive, it is called an x drive.

7 Likes

I would like to add that many swerve drive designs perform great in FRC, where you could CNC custom gears and support structures.

I’ve been watching for a while amazing CADs and design ideas that Kyle had been putting out lately, but I have to agree with OP that for VRC they are unrealistic due to the extra complexity and friction when attempting to build them from VRC legal parts. The price you pay in the power, lost to the friction, is just too high to justify swerve drive at this point over the traditional meta tank and x-drives.

However, it doesn’t mean that there is no point to keep researching new drive train ideas. I am very optimistic that innovation could still happen, if you could manage to reduce the complexity. For example, both omni and mecanum wheels were themselves literally the re-inventions of the wheel.

13 Likes

To respond to some comments:

Yes, @Dimension_360, I understand there are middle schoolers here and I apologize for not mentioning them. However, to my understanding I don’t think middle schoolers are responsible for alot of the drama that has been going on in here.

As for your comment about my harshness, I agree, but I also think it needed to be said. We need to act like adults.

For others - @technik3k’s information is very helpful as well, and he has linked some more in depth resources on x-drives. He also talks a bit about mecanum drives, which I did not mention because they weren’t apart of the original discussion (though they should’ve been in retrospective).

@9MotorGang There are a variety of names for X-Drives/Holonomic drives that people come up with, but they all refer to the same thing. I’m not sure where “Bendy drive” came from (though I have a strong suspicion), but it’s just people messing around.

14 Likes

I’m a big fan of swerve drives.
Here are some of my favorite reveals and uses of them in FRC!

As Mr. Brendan mentioned, swerve drives are very useful because they have the ability to utilize the full power of their motors in any direction, because the motors turn the wheels themselves (much like a shopping cart’s front two wheels can do) to face the direction you want to go. Another benefit of swerve drives is that they can change into an angled “bendy drive” (or x drive, for you other folk), and reach speeds approaching infinity (a topic thoroughly fleshed out in the vtow discord) with a proportional loss in torque, of course. While this is neat, the lack of torque leads to motor burning out, making it unviable to go this fast, as Mr. Taran mentioned. Even so, swerve drives are a very cool type of drive, and I’d love to see them this year.

While many on this forum think it is unviable, or even impossible, to create and use one practically in vrc, I disagree.

As you all saw in the other swerve post that was made today (which unfortunately got taken down), Kyle from 8181x created a swerve drive that used a differential to allow one motor to both power the wheel’s rotation about the axis and rotate the wheel itself.

Personally, I’m rather surprised that swerve drives have not been used more frequently in vexu. While I am not a part of vexu yet, it seems that with the capability of custom machining parts, one could create a swerve drive with much less work that someone in vrc for instance. Vexu teams have the ability to print ring bevel gears, which look like this:
image

I suspect that as more vexu teams (and the new vex ai teams) learn of how to use computer aided custom manufacturing and 3D modeling, we will see these types of drives very soon.

Another note on the possibility of swerve drives is the fact that it may be viable to do a 2 pod swerve drive, which would work very similarly to a shopping cart like I mentioned earlier. 2 of the wheels would be powered (or unpowered) in the back, and 2 wheels in the front would be able to work like a swerve drive with 2 motors powering each wheel.

I believe that with good building techniques, a low friction system can be made that will allow for this form of drive to be relevant one day in vrc. It will undoubtedly be harder than in vexu, but I think it can be done.

Thank you Mr. Brendan for making this post.

14 Likes

Technik is correct - Swerve drives are absolutely more viable in other competition systems (notably FRC) due to their differing rulesets. I refrained from posting any examples of mechanisms from other competitions because I felt it would set unrealistic performance expectations for those trying to replicate them in VRC.

6 Likes

Jumping in here. My former FRC team 1640, did a ton of prototype work on swerve drive using VEX parts.

We also built the famous “Twitch Drive” that moved the wheels 90 degrees in about 2 seconds.

People hate swerve, but Team 1640 has done years of research on it for FRC and has some pretty sweet modules they make.

Nice to see this thread!

10 Likes

Kyle asked me to post this for him since he doesn’t have an account:

A feature I feel not given justice provided by swerve drives with omni wheels is the ability to change mechanical advantage by the degree in which wheels are angled towards each other. Ethan touched on that you can reach theoretically infinite speeds, but I’d like to talk a bit more on how this is beneficial.

Note: in the OP it was said “And while there are arguments for variable speed and torque, those things can be somewhat achieved using A) programming or B) gearing”. This isn’t necessarily true, as in programming you can’t change mechanical advantage (just pump in excess voltage at lower speeds, which is what you should be do regardless of drive type, however stall torque is the same), and there isn’t any other 4 motor holonomic drive that can change mechanical advantage with gearing, only tank (which I think can be low friction enough to be viable).

Being able to have variable mechanical advantage provides two main benefits: Being able to have a “pushing mode” (where your wheels are on tank style) and a “speed mode” (where your wheels are angled against each other), and having a quicker acceleration time. Both of these benefits cause a change in your drive speed: having a pushing mode means you don’t need as much torque on speed mode -> faster top speed, being able to reach top speed quicker than necessary -> faster top speed. This also helps counteract some of the side effects of the overwhelming friction, which would reduce your pushing power and acceleration time.

image

Here’s an illustration of the the torque levels of a swerve drive (green) compared to a normal drive (red). The x axis is the drive speed in rpm of a 4.125" wheel equivalent. The y axis is the torque exerted times the mechanical advantage. This is assuming a 4 motor drive with a 180 rpm speed equivalent in tank mode and a 45 degree angle in speed mode (since 45 degrees has the least roller friction, although you could go faster, as Ethan mentioned). As you can see, there is more torque in the swerve drive for most speeds, which contributes to acceleration. Between 180 rpm & ~250 rpm you can see the gap in torque decreases as angle would change gradually (with the acceleration of the bot) to the final speed setting.

I’d certainly agree that my design as of now has to much friction to be viable, but I think with changes and friction reducing methods (you can certainly reduce the quantity of top gears and take out the encoder, 3 pods/6m has potential too- variable MA is messed up with it though), the vex community has the potential to develop swerve drive as the strongest and most versatile vex drive. Plus swerve drive bots look cool, and that’s the determining factor for a good bot :).

8 Likes

I am going to throw a really crazy point in the ring and mention that a 3-wheel swerve drive only takes 6 motors (at the cost of a more complex and less-stable base).

Swerve drives, although not often practical given VRC motor limits, are a very interesting drive design. They provide theoretically higher drive efficiency at the cost of significant complexity. Engineering is all about trade-offs, so good luck!

If you find a way to use them effectively, many of us would be quite interested in seeing how you addressed these trade-offs. Just because it isn’t often done does not mean it can’t be done.

Just remember to learn and have fun!

4 Likes

I would like to note, since I don’t think it’s been mentioned that Kyle also has a diff swerve for 3 wheels:

(Credit of course to Kyle from 81818X)

8 Likes

The only thing that I see with this is that an x-drive just seems like a much better drive logistically. A lot of teams gear up their drive already, and since speed and torque are inversely proportional, there’s no difference between gearing up a tank drive or making an x drive. Swerve drives also only use half of the motor power of the motors allocated to the drive (uses 6 motors but only 3 are pushing other robots). Also, there are only two motors available for any other mechanisms which is almost never enough. Of course you could go ahead and only use two swerve modules and do dummy wheels on corners but that’s only 2 motors of pushing power. Doing a 393 swerve allows more motors but 393 is just so weak compared to v5 that it just seems like a v5 x drive, heck, even a v5 strafe wheel is a better option.

As you said, it would be interesting to see but I don’t think we’re gonna see it. It’s plausible in frc because the only limits are (pretty much) weight, cost, and size. The swerve drive is much better than an x-drive, to the point where it’s called a “poor mans swerve” in frc. The only thing that’s stopping people is motor limit and power

This is not true. With a differential, you can utilize the full power of all the motors and still have 4 left over for other mechanisms

2 Likes

I explained it afterward cause ik that I explained it badly. They only use half the motor power of the motors allocated to the system. 1 swerve module uses 2 motors but only has one motor of pushing power

That’s still not true for a diffy swerve drive. You retain full power of all the motors when turning the wheels themselves about their axis and rotating the shaft when using a 4 wheel 4 motor differential swerve drive

2 Likes