Mechanum Wheel Sizes

[nxtsoccer]

Hello, my team (4087) recently completed our first FIRST season at FRC. We had a blast and are talking a lot about expanding the size of our team, budget, and quality of our robot for the 2013 season. Anyway, one major topic that my team has been talking about is holonomic drives, specifically mecanum ones for next season's robot. I've been doing some research into these wheels and have found many mixed success stories. At any rate, I found much info online as well as on Chief Delphi, however, I could not find a good thread discussing which sizes of mecanum wheels generally work best. I read somewhere that teams had problems with the 6" AndyMark mecanum wheels deforming and the 10" AndyMark wheels seem very expensive, too large, and too heavy for practical use in FRC. Does anyone have any first-hand experience as to which size(s) work well (movement not too bumpy, resistant to damage, ease of mounting, etc.)? Any recommendations and/or comments are welcome! Thanks in advance!

[EricDrost]

Not to be a downer but in most FRC applications a standard 6 wheel "drop-center" tank drive with grippy wheels like Colsons will bring you far better performance than a mecanum drive train.

However, if your team is dead-set on using an omnidirectional drive train, I cannot stress enough how important it is to build the system you want to use in the OFFSEASON. This way you can work out all the kinks and learn from your mistakes. Building an omnidirectional drive system for the first time during build season will bring unexpected challenges that will require much more time than you think in such a short 6 week build window.

As for wheel size... I'm by no means an expert but the better applications of mecanum I've seen tend to be using 6" wheels.

[CalTran]

Originally Posted by EricDrost

Not to be a downer but in most FRC applications a standard 6 wheel "drop-center" tank drive with grippy wheels like Colsons will bring you far better performance than a mecanum drive train.

I beg to differ on this point. The simple reason why tank usually outperforms mecanum drives is because drivers usually don't use mecanum to it's fullest potential. Mecanum drive is an unconventional drivetrain, thus it takes practice more than anything to be able to use the strafing regularly in your driving. People generally are not used to having that ability when driving, nor are they used to being able to spin on a point. The biggest thing you need if you're going mecanum is practice using it.

Aside from that, everything else Eric said is true about mecanum, though our team managed it in season. More complicated omnidirectional drives (ie swerve or octacanum) would require much more time.

[Gray Adams]

Originally Posted by CalTran

I beg to differ on this point. The simple reason why tank usually outperforms mecanum drives is because drivers usually don't use mecanum to it's fullest potential.

Even at its fullest potential, I'm not convinced you will get better performance our of a mecanum drive.

I can't speak about mecanum wheels specifically, but we've come to like using the smallest wheels we can. They have a lower moment, they're lighter, and you need less gearing.

[Djur]

This thread has quite a long discussion about mecanum wheels. Take a look at it.

[slijin]

I wouldn't be so quick to dismiss mecanums, but there are better ways to achieve maneuverability - of course, that's another discussion itself.

We used 6" mecanums in 2011 with great success; we had absolutely no problem whatsoever with wheel damage. We did come away from that season with two particularly important lessons, though:

1 - make sure your wheels are always in contact with the ground. Although it seems obvious as the kinematics of mecanum drive rely on all 4 wheels, it's not obvious when your chassis is warped. Following unaccountably erratic driving control halfway through qualifications at CMP last year, we discovered slight chassis warp, on the order of 1/8" to 3/8". It ended up with two mentors jumping up and down on our chassis to fix the warp (we students weren't heavy enough ). After that, driving control was much more precise.

2 - precise speed control is a must. Because the nature of mecanum requires that each wheel be individually controlled to a relatively high level of precision to enable accurate driving, it's imperative that the software is capable of precisely controlling wheel speeds relative to one another. Unfortunately, it wasn't as easy as tossing CIMs into Nanos and bolting them to our chassis. Developing effective speed control took our software engineering team a good amount of time, and it was only thanks to that that we were able to achieve the level of driver control that we did.

[IndySam]

AM sells 6" 8" and 10" mecanum wheels.

The 6" and 8" are geared towards FRC use. We have used 6" HD wheels and loved them. We never had a single failure.

10" are pretty cool but a bit of overkill for FRC.

[Kris Verdeyen]

It's perfectly natural for a team your age to feel things like this. Many of us have experimented with holonomic drive trains at one time or another. It's nothing to be ashamed of, and it's good that you've come to us for help. With hard work, focus and determination, you'll come through this, better than ever.

The thing to remember is this: every single person on every single team, including the best teams in FIRST, have, when first presented with them, looked at mechanum drive trains, and said, "Whoa, cool!", and not one robot built with them has made it to Einstein in at least 2 years (because that's how far back I've checked).

It's not because they suck inherently, it's because it's hard to make them run to their full potential. So, because you'll ignore this warning and build it anyway (because, "Whoa, cool!"), the best advice I can give you is this:

1. Encoders on all 4 wheels
2. Only run it closed loop (typically velocity mode)
3. Make sure each one of the (48) little rollers is free spinning before each run. Bent wheels can cause these to bind.
4. Understand what the code needs to make the math work, especially any code you don't write yourself.
5. It still probably won't work great on anything but a flat floor.

People below me will contradict points 1,2,5, and maybe 4. They will have counterexamples. They also probably have a grandfather who smoked 4 packs a day and lived to be 100. Ignore them, believe me.

[nxtsoccer]

Originally Posted by Kris Verdeyen

1. Encoders on all 4 wheels
2. Only run it closed loop (typically velocity mode)
3. Make sure each one of the (48) little rollers is free spinning before each run. Bent wheels can cause these to bind.
4. Understand what the code needs to make the math work, especially any code you don't write yourself.
5. It still probably won't work great on anything but a flat floor.

Woah! First off, I'd like to thank you all for the amazingly immediate responses! Second, based on recent posts, I'd like to put something out here: we're a one-season team with only two or so of our members having any previous robotic experiences (myself and one or two others). Additionally, this is the first time I (and I believe my teammates) have gone beyond the LEGO Mindstorms NXT (r) level of robotics. In a nutshell: I''m really not sure what to think of some of these posts, much less what they mean! I know what an encoder is (measures/counts rotations), but I don't see how that would be beneficial to a holonomic drive. As for code, if we decide to go with mecanum wheels, they'll almost certainly be driven using four jaguars and the code for holonomic drive found in the WPI Robotics Library in LabView, without any changes.

Thanks for understanding our current lack of experience in pretty much everything at this point!

[Grim Tuesday]

357 doesn't have a big presence on Chief Delphi, but if anyone is going to get to Einstein with mecanums, it will be them. I'd suggest trying to get in contact with them if you're looking for help designing driver practice for this or any other holomic drivetrain.

[KrazyCarl92]

Originally Posted by nxtsoccer

I know what an encoder is (measures/counts rotations), but I don't see how that would be beneficial to a holonomic drive.

Encoders are beneficial to a holonomic drive because these drives largely rely on the kinematics involved with the wheels going at very precise speeds. Since an encoder can measure the revolution (in this case of the drive axle or an axle coming off of the gearbox), it is the proper sensor to close that loop and accurately control the speeds of all of your wheels.

If a mecanum drive is coded for voltage control (no closed loop, just set the motors to a given voltage), you could have a situation where your driver commands the robot to drive straight forward, but only 3 of the 4 wheels are driving close enough to the same speed. That fourth wheel driving at a different speed than the other three will cause a slight rotation/strafe in the driving, resulting in a robot that will not be able to drive in a straight line. This is why speed control of the wheels is very important and encoders can be used to achieve this.

Good luck, and I hope you guys have success with your off season projects! It's this kind of work that will prepare your team for a successful season.

[rocknthehawk]

If it's something your team wants to do, got for it! There will be naysayers that will give every reason not to try it, but what fun is that? Pushing yourselves to try something new, and learning in the process, can be hugely beneficial.

This was our (Team 126's) first year using mecanum wheels. I don't think anyone would have changed this decision. Our camera alignment was fantastic, with the ability to strafe. We spent a lot of time calculating gearing. We ended up with a final drive of about 17.5:1, CIM to wheel speed. Secondly, lots of drivers practice. Usually 4 days a week for 2 hours or more.

We did not run encoders on our robot. We initially made custom mounts for encoders, but ended up finding we did not need them. I don't know the programming side well enough to understand why, or what we did differently. Just that it didn't operate any differently for us not using them.

We used 8" AM wheels. They are heavy. I don't think the washer and brass bushing for the rollers is the best design possible, but it's simple and cost effective.

[nxtsoccer]

Hey, do you guys think we'll need this: http://www.andymark.com/ProductDetai...Code=am%2D0279

It's a hex-bore ball bearing to fit the shaft of the toughbox nano, probably would be mounted on the other side of the shaft. On that note, are the toughbox nano's built-in bearings strong enough to support the weight of the robot, or would they need to be supported by bearings on both sides?

Eh, I think it's a confusing post.... Basically, how many (if any) bearings will we need other than the built-in ones to support the robot's weight without breaking something or unnecessary wear and tear? Any ideas?

[Nemo]

Originally Posted by nxtsoccer

Hey, do you guys think we'll need this: http://www.andymark.com/ProductDetai...Code=am%2D0279

It's a hex-bore ball bearing to fit the shaft of the toughbox nano, probably would be mounted on the other side of the shaft. On that note, are the toughbox nano's built-in bearings strong enough to support the weight of the robot, or would they need to be supported by bearings on both sides?

Eh, I think it's a confusing post.... Basically, how many (if any) bearings will we need other than the built-in ones to support the robot's weight without breaking something or unnecessary wear and tear? Any ideas?

You can go either way. In 2011 we bought the long axle versions and supported the opposite end of the shaft with simple blocks of plastic with 0.5" holes bored in them (we used the keyed versions). We liked that a lot. The disadvantage of doing that is that it's harder to get a wheel off if you need to replace it or repair it. Plus it is extra work to build.

The Toughbox Nano is strong enough to support the robot's weight. Of course, you want to put the wheel as close to the gearbox as is reasonably possible to minimize forces that would try to bend your axles.

[Taylor]

In 2011, we used a drive base very similar to what Scott has described. We then mounted a modified C-Base to the top. We had a nice surface to mount our electronics, and we were careful to leave them easy to reach.

http://www.chiefdelphi.com/media/photos/36120

It was the best, most reliable drive train we've had in 8 seasons, and if the game calls for it again, we'll certainly look at it again in the future. It was built in just a couple hours.

I don't know if this exact setup would work on a non-flat playing field.