Mecanum + Omni Drivetrain Pros and Cons

Hello everyone,
As #6838, we have used mecanum wheels for 3 years in a row right now and really enjoy the maneuverability it provides. Our driver really started to master mecanums in order to quickly rotate around defenses and quickly aligning sideways. So, we are trying to improve our mecanum experience with a new drivetrain design.

We are planning to have 4 regular 6" mecanum wheels on all four sides and 2 6" omni wheels in between mecanum wheels. You can see a scheme below. ("+" resembles mecanum and “*” resembles omni wheels.)

 __________
+|        |+
 |        |
*|        |*
 |        |
+|________|+

To begin with, I am almost certain that I read about this but we could not find any posts or examples of such a drivetrain. Could you guys tell us about the advantages and disadvantages of such a design and tell us your best practices if you saw or built such a drivetrain.

(Without being very sure, I am planning to have an arcade drive for omnis and a regular cartesian drive for mecanums at the same time. I am a bit concerned if this would damage mecanums as mecanums will move forward at a slower rate as they lose their potential power due to angled rollers. What do you guys think about running arcade + cartesian? )

1 Like

My first question would be: Why? I don’t see any advantages to designing a drivetrain like this. It would still drive like a mecanum robot so why have the extra weight in the middle for omni wheels? Also, if you have mecanum wheels there is absolutely no reason to drive it like a traditional tank style drivetrain. You lose all the advantages of the mecanum drivebase.

Just pick one

2 Likes

We will be using it to get encoder reading from the omnis for auto which probably will give us more accurate data as mecanums usually slip. We will still be running driveCartesian() which will give us all the mecanum maneuverability. In addition we will be running omnis which we think will give us a faster and more defense-resistand drivetrain. I wanted to know about how this theorical drivetrain would turn out practically.

The omnis won’t help against defense. If y’all are set on a mecanum drive train I would look at 2017 1986, because they used a dropped Omni to get distance without making it a part of their drivetrain.

Otherwise, I would highly recommend looking into standard west coast.

You can get pretty accurate encoder data from mecanums if you are careful. However, it will require much more work than a standard WCD. Practically, it would be a mecaum drivebase with no differences.

I believe you mean 1986 Team Titanium, not 1987? 1987 used a tank that year it looks like.

3 Likes

I’m a big advocate for mecanum drive and even I don’t see this as advantageous. You really won’t get much extra benefit from the center omnis and you’re trading a lot to put them in. Besides what others have said, there’s the weight problem. Mecanum already comes at a weight cost, since you need four motor/gearbox sets to run it. With this you’re adding in two more full sets (one for each omni) so you’re upping your motor/gearbox weight by 50%, when that already a big part of a mecanum drive train’s weight to begin with. I just don’t see the benefit. If you’re going to use mecanum (and our team does, quite successfully) you just have to accept the trade-offs that come with it and design other aspects of the robot to compensate for and/or take advantage of what mecanum is.

1 Like

Thanks for the advice, we thought it would help us a lot with speed and defense. However, your experience tells exactly the opposite I suppose :confused: . Thanks anyways. Also, I was wondering how you guys are managing auto with mecanums. Can you tell me about your auto routine? I would really like to talk to your programmers if you guys are okay with that.

My gut instinct is to echo folks here - you might get a bit more traction with the ground in the fwd/rev direction, but you’ll still be easily pushed from the side. This means you’ve added complexity, but still not mitigated the usual “mechanum is easily pushed around” concern.

That being said, I’m all for smart innovation. If you have time in the offseason, it would be a cool experiment to try a bunch of different wheel setups (mechanums-only, what you have above, Titanium-style double mechanums, etc.) - measure the max pulling force in all directions each type can exert, and use the data in the following season to inform the design decision.

This is only if the playing surface is perfectly flat and your chassis is also able to put even pressure on the 4 meccanum wheels. When there is uneven pressure on the meccanum wheels, their function becomes unpredictable. Adding the two omni wheels would take some pressure off the meccanum wheels, otherwise they do nothing.

1 Like

Check out how FTC does this, they are often the king of Mecanum. Often, rather than an addition wheel as part of the drive train, they use spring loaded omnis that push on the surface the ground. These omnis are unpowered and attached to an encoder. 1986 did this method in 2017 (can see at 84 seconds).

I highly suggest doing using spring loaded omnis over the other design. Your team can still design something new without sacrificing an already solid design. Worst that happens if the omnis don’t work? You play exactly like you have in the past.

https://www.youtube.com/watch?v=cinJ-xmWA1o

What you’re proposing has been done at least in Vex, which they call asterisk drive. In this video they use a killough drive instead of mecanum in the corners, but it’s the same basic concept.

@burakdemirelli This post is important. Also, if you are using omnis with encoders, remember that mecanum drives drift, meaning you will have some sideways motion not measured by the left and right encoders you show in the diagram. I recommend adding an omni in front or back perpendicular to the other omnis to measures this sideways motion as well as not having the omnis powered at all.

I can put you in touch with our programmers, but they haven’t had any experience with doing auto with mecanums. They were all new last year and we drove by camera during sandstorm rather than try an auto routine. All our experienced programmers have graduated and I’d be hard pressed to get you in touch with them. We’ll likely be doing some work on auto this year, so the current team will certainly share what they know and learn. In general, it’s not really that different from any other auto programming in my experience, though I am not a programmer myself and can only tell you what I’ve seen our past programmers do.

I actually have built a drive train with four mecanum and two omni wheels, but the omnis weren’t normally in contact with the carpet. It’s built on the AM14U4, with 6" mecanum wheels. I noticed that the ground clearance was pretty short, especially under the mount plate for the standard central gearbox, and I was concerned about it high centering on a platform such as 2018 and 2019… As such, I put 4" omnis on those shafts, driven off the same gearbox as the front wheels, but appropriately faster. This chassis does NOT have encoders; if I were planning an auto routine with it I would make certain to drive all four mecanum wheels forward at the same speed when entering/exiting a platform.

Here’s a link to a thread about this very thing, I made a basic CAD model of it.
=> Mecanum Chassis Idea <=

1 Like