Mecanum vs Tank Drive & Gearbox/Wheel Selection

Technical Technical Discussion
1

This week (similar to most teams I’m sure) has been full of design arguments & debates (healthy & non-healthy). After a full week of 5 hour days, we now have a mostly completed design. However, there is one major subsystem of our robot that we cannot decide on: the drive train.

I’ve decided to create this thread to present the pros/cons of both mecanum and “regular” drive trains that we have debated for the past few days. I’d love to hear your thoughts / other points you have to offer and plan on using this data to make a final decision.

Pros of Tank:

  1. More traction
  2. Faster acceleration / top speed
  3. Better for defending
  4. Cheaper (however, expenses are not weighed heavily on our overall decision)
  5. Lighter

Cons of Tank:

  1. Less maneuverable / harder to be precise
  2. More susceptible to tipping if pushed from the side
  3. Harder to drive/handle (in my opinion)

Pros of Mecanum:

  1. More maneuverable / more precise
  2. Less susceptible to tipping from the side
  3. Easier to drive/handle

Cons of Mecanum:

  1. Heavier
  2. Expensive
  3. Takes up more space
  4. Can easily be pushed around
    ^ the mechanical lead & I both feel as though this could lead to issues with being pushed into the opposing alliance’s null zone.
  5. As stated by reddit.com/u/bmar1257: “Just a warning for everyone. Mecanum results in a significant penalty on many teams’ picklists.”
    I’ve also seen this first hand, with a team scouting a team I was helping. In the middle of scouting the team, they left after they were told that the team used mecanum wheels.

On top of this debate, we have brought shifting gearboxes into the equation. We feel as though having defensive capabilities (that is if it does not significantly affect our offensive capabilities) is very important this year.

Therefore, I, and a few other members on our team, believe that a shifting gearbox will be very useful when it comes to the ability to push robots into our null zone, cover choke points, and prevent ourselves from being pushed around.

However, our mentors have brought up some useful points when deciding on whether or not this is necessary.

  1. If we use grippy wheels, (such as plaction wheels) will we even need a higher gear ratio?
  2. Are the benefits of a shifting gearbox worth the added weight / power consumption / area coverage?

We have also talked about whether or not we should use West Coast Drive. From my understanding, the wheels are just placed on the outside of the chassis, so I’m not quite sure what the pros/cons are.
Side note: none of us know much about drive trains other than mecanum / regular tank drive.

Thank you all very much for reading!

UPDATE
Some of the replies have mentioned the level of difficulty with coding / driving.

I should have mentioned that I am the driver, and we used mecanum drive last year.
Here is a video of one of our matches: https://www.youtube.com/watch?v=N_dYvD2Zzcc
^ red alliance, 2 green leds, 6077.

2

Totally ignoring the rest of the thread real fast. If they left basically right after hearing they used mecanums, thats kinda un-gp. I can understand the reasoning, but theres way better ways to of done that.

The difference according to the most recent CD thread I read on WCD’s is that WCD drivetrains have a “live axle” wheel direct from the gearbox as opposed to being all on dead axles for driving. Someone more knowledgeable will probably correct me though, so take above with a grain of salt.

Personally, id recommend Colson or KoP wheels, mainly because of the extra hassle mecanums are if you never done it before (extra coding, encoders, and more attention to CoG) and if you usually dont allot much driver practice. These are essential to making a mecanum drive work well. Check out 492’s 2017 bot for inspiration if you decide to use mecanum.

3

I always see this come up in discussions about Mecanum but I believe that the programming difficulty is exaggerated. We used the AM Nanotubes with tank drive in 2017, which happen to have 4 seperate gearboxes. As an offseason project, we converted it to Mecanum and it took about an hour to make it work, with field oriented drive. Most of that time was spent getting the joystick working correctly and switching the motor controller ports around because the PWMs got plugged in the wrong order.

I agree with encoder issues and attention to CoG though.

Getting it to drive straight took a little bit more time, since the weight distribution was not at all ideal (all of the weight was at the back near the climber so we didn’t swing). But this was also correctable using the gyro, and it seemed to work pretty well. Alternatively you can design your robot to be weight balanced.

As far as encoders: you pretty much can’t use them because there is too much slip with the rollers. The wheels can stop turning but the rollers can continue, moving the robot without moving the encoders. Most teams opt to add omni-wheels that are undriven, one parallel to the sides of the robot and one perpendicular, then connect encoders to those instead. This way you can get a more accurate estimation of the distance the robot has actually traveled.

WPILib has full support for Mecanum drive now, so it’s really easy to get started at the least. No comment on Mecanum as a design choice though. It’s a choice you make with its pros and cons as you’ve already established. I am personally a fan of a tried-and-true tank drive, but there are definitely games where Mecanum can give you an advantage. If your concern is alignment, the scoring platforms this year were 3ftx4ft. With a little bit of practice, I had no issue lining up our tank drive with the gear peg last year which was a much smaller target to hit.

Shifting gearboxes - a whole 'nother can of worms. I would say that, just like Mecanum, certain games warrant their use. A great example would be 2014, which was a completely open field. High gear could be used to sprint across the field and outrun defense, while low gear could be used to push through de or even be the de. I would personally say that, at this stage, you should build whatever you can get done the fastest. If you already have single speed gearboxes in house, use those. You want your drivetrain done ASAP. If you’ve already got shifters in a good reduction, then it might be worth looking in to.

4

I agree, what that team did was not right by any means, I just used it as an example.

Also, I should have mentioned that we used mecanum wheels last year, and I’m the driver.

here’s one of our matches: https://www.youtube.com/watch?v=N_dYvD2Zzcc
2 green led rings, red alliance, 6077.

5

My team uses a six wheel rocker tank every year with a shifting gearbox. It has worked very well for us.
A rocker tank with a small rock such as 0.09 would improve maneuverability at the cost of some stability. Using an 8 wheel rocking drive train would give more stability at the cost of weight. However, rocking does require more work and precision in terms of machining.
I would highly recommend a shifting gearbox. They are great. We’ve used vex ball shifting ones in the past and built our own.
In terms of mecanums, I don’t think it’s the best drive train for this year. I can see a lot of defense being played and mecanum drive trains tend to be pushed around.

West coast vs normal tank:
Both have double supported axles, it just depends where the second support is. In tank it is support, power transmission(ie chain,belt) wheel, support; west coast is support, power transmission, support, wheel.
For normal tank my team uses axles with a quarter through hole and use a bolt to hold it in place. This allows us to easily remove the wheel, where as on west coast you just slip it off the end. West coast uses live axles, which mean they turn with the power transmission and wheel and need to be supported by bearings. It’s great for encoders.

You could also considering doing a tank with a combination of omni and traction-y wheels.

Colsons are great wheels. Good Luck!

6

  1. We were able to have consistent side peg and center peg autons using only vision tracking, a gyroscope, and timings last year. In terms of how we drove straight, I had a button on the side of my joystick for driving straight forward and sideways (gyro corrected). There’s a link of this at the bottom of the initial post.

  2. You’ve stated that shifting gear boxes have their time and place, do you think this year is one of them?

  • If so: do you think the few day wait will be worth the benefit it provides?
  • Also, would I want my drive train done asap just for practice, or for auton writing? I suppose I don’t see how I could do much with auton considering we don’t have a field(ish) setup yet, aswell as weight on the chassis.
7

  1. Since you guys have had good experiences with the programming and driving of Mecanum, I guess I didn’t need to type out that huge post :stuck_out_tongue: That leaves the option on the table for sure.

  2. I am just hesitant to make a decision based on a fairly limited amount of information, and I also don’t want to be making a decision for you. I will tell you that 3082 is using Vex 3 CIM ball shifters with either 2+1 or 1+2 CIM and MiniCIM (leaning to 1+2). We made this decision BEFORE the start of the season and ordered the gearboxes, which arrived a week before kickoff. After last season, we were fairly comfortable with using pneumatics so it seemed like a logical step. We have planned it out quite well and will hopefully be driving tomorrow afternoon.

I personally think that there is value in being able to use the low gear (ours is REALLY low) to make fine adjustments while placing blocks on the scoring platform(s). I expect to spend most of the time in high gear (~16fps) and only shift to our 5 fps low gear when it’s absolutely necessary. I am not envisioning as much robot-to-robot defense as last year simply due to the presence of the Null Territory, which is probably where a lot of defense would have been there. If I am mistaken, and there is more defense, the low gear option is of course there.

Do I think that a single speed team will be at a significant disadvantage to a two speed team? Absolutely not.

I am not 100% sure on the amount of time it’ll take you to buy shifting gearboxes. The 3CIM ballshifters, 3 stage version are out of stock at the moment, limiting you to either 3CIM, 2 stage (only really works with 4inch wheels) or 2 CIMs (which is perfectly fine). I think that it could be worth it if you can handle the design and build of a West Coast drive or similar drive train.

Do you have a kit chassis? If so, the AndyMark Evo shifters (NOT REDLINE) were designed specifically for that. However they are a bit pricey. That will also allow you to get the drivetrain done quickly.

I am just going to say one more time that there is no “better” option. Single speed and two speed are both great options. Look at last year - 254, 971, 973, 1678, etc etc etc all had single speed drive trains (I think, please correct me if I’m wrong). You need to decide, in the context of your team’s resources (people, time, equipment, etc) and the fact that we have less than 40 days left of build in order to decide which is the best option for you. I hope I have provided enough information to help you decide. I just cannot make that decision without knowing 6077.

For getting your drivetrain done ASAP: your drivetrain is almost always the most important mechanism of the robot. You want it to be done because everything else goes on top of it. You can get some driver practice in by attaching deadweight (securely) to places where you think weight will be distributed. Obviously that’s an approximation but it helps a lot. Autonomous development is helpful. But I firmly believe that a robot that cannot move cannot be competitive in Power Up. Once you get your drivetrain done, you can focus on better things. Want to climb? Want to place 800 cubes in the scale during auto? Cool, but you need to move first. But mostly, you want to know that your robot can move.

If you’re building your drivetrain until week 6, there’s not much good that can happen if/when something goes wrong during those last few days. Trust me, my rookie year in 2016 was not a good one for our drivetrain. The earlier you mess up, the more time you have to fix it.

8

To be honest, using the low gear for precision was not what I had in mind. What’s wrong with lowering the power output of the motors for fine adjustments?

Also, is using a shifting gear box really that big of a commitment? I feel as though it wouldn’t take more than 30 minutes - an hour (although I could be wrong) and 2-3ish people to get it setup correctly. I suppose the only factors would be money, which we should be good for, and shipping time, which is important like you said.

I suppose one thing I should add is we will be placing our first AndyMark order on Sunday. Meaning we probably won’t start building / fabricating until late next week anyways.

Thanks for the great info btw!

9

As a rookie team who used a shifting gearbox (Andymark EVO-Shifter) last year, I think I’m qualified to answer some of your questions.

  • What’s wrong with lowering the power output of the motors? There happens to be a minimum amount of power required to overcome static friction. This amount of power increases when mechanical advantage decreases. Essentially, you end up with a higher minimum speed for your robot. This is rather annoying when trying to turn, say, 5 degrees. You get the robot to just barely move and then BAM! you’ve turned 15 degrees.

  • Commitment? Absolutely. Setting it up is the easy part, and that took my team an entire build day (9-5). Money is also a factor. However, the biggest (and most overlooked) cost is the increased need for driver practice / software implementation. You can either train the driver to shift in the correct situations, or you can automate the shifting process through the program. In addition to all of that, shifting means you are committing to using pnuematics (servo shifting is terrible), which adds complexity to the control system. It’s doable, but it is a commitment.

Don’t let all of that discourage you though. It’s definitely possible - my team did it.

10
  1. That wasn’t really a problem for me last year, although we were using mecanum wheels so I’m sure that had something to do with it.

So I’ll assume that it’s much harder to make tiny adjustments with a regular tank drive setup, but I’m not sure that making small precise turns will be that big of a deal this year.

  1. We are already using pneumatics for our climbing mechanism, so that wouldn’t be much of a problem. Also, how would you automate the shifting process? Not asking how you would actually write the code, that shouldn’t be much of an issue, but how would that work logistically?

Thanks for your help :)))

11

I haven’t actually automated it myself, but my thoughts are: figure out what the max encoder velocity is for low gear. Shift to high gear when the velocity is above, say 90% of max. Shift back to low gear if it drops to say 80%. Note the gray area - thst is intentional, to avoid rapid shifting. The general idea is you can accelerate faster in low gear, so starting in it makes sense. Once you have used all the acceleration, move to a higher gear for greater maximum velocity.

12

Dear Jeremy Styma + Team# 6077,
Our team 6680 is trying out Mecanum Wheels for the first time and were wondering if we could see some wiring designs, how you programmed wheels, etc. We are excited to use these wheels, we just want to be on the safe side on assembling/programming them. Thank you!

13

Dear Jeremy Styma + Team# 6077,
Our team 6680 is trying out Mecanum Wheels for the first time and were wondering if we could see some wiring designs, how you programmed wheels, etc. We are excited to use these wheels, we just want to be on the safe side on assembling/programming them. Thank you! :]

14

If you are not going to program mecanum as field-centric, do not use mecanum. Make an H-drive instead.

An H-drive has all the same advantages as a robot-centric mecanum drive and fewer disadvantages. It can actually PUSH for one of them. It also won’t get you heavily docked on some teams pick lists!

Mecanum should be thought of as swerve-lite, if you aren’t going to program it’s movements to mimic a swerve you are leaving a large number of possible advantages on the table while keeping all of the cons.

15

Team 852 used mecanum last year. We found that while it was more maneuverable, we lost a lot of pushing power and thus got pushed around. This year we are going back to tank drive.