Another Drivetrain Idea - "True Octanum"

[SpeedFreed]

While reading the recent thread about "Box drive", I was struck with the idea of that configuration made out of mecanum wheels. Now this is pretty much just a thought experiment, but I got to thinking that it does solve one of the biggest problems mecanum has, which is the loss of power in strafing. Here is a concept drawing:



Notice the 8 mecanum wheels, making this, in my option, more worthy of the "octanum" name.

In an environment with no unwanted friction, a typical mecanum configuration can reach full speeds forwards as well strafing. For the longest time I thought that mathematically mecanum were limited to about 70% speed in strafing action, but in actuality the 30% lost entirely to entropy (booooooo).

However, as shown above, this configuration has the second set of mecanum wheels that will be able to strafe at the same speed the other set can go forward. Since mecanum wheels are not mathematically limited, the set not being "used" should just go along with the ride, and while not providing much use in moving that direction, they shouldn't drag much on the ground.

Each pair of mecanum wheels (two per corner) should be able to be linked via a clever use of bevel gears, and the driving/coding should be no different than typical mecanum.

I have only been able to think it through forwards, backwards, strafing, turning, and diagonals, and as far as I can tell I should work. Though I can't tell when it comes to complex movements.

This should allow you to make a completely symmetrical, omni-directional drivetrain; the only other ones being swerve and holonomic. It also should reduce mecanum bounce, as well as increase mecanum's traction some (though that is a complete guess). Furthermore, you get the advantage of full motor usage in the primary direction, which you lose in something like H drive (which has to split up the motors for different directions).

What do you guys think?

Note: Again, we aren't planning to actually do this. In fact we have already designed built and are now testing the other "octanum" for next year.

[AustinH]

How do you ensure coplanar ground contact between all the wheels?

[kitare102]

Looks like you might need to go into the realm of spring suspension if you're worried about coplanar contact. If you made the corner unit as a whole independent from the main frame, I think that would be a functionally sound way of pushing the mecs down into the floor surface. Obviously that would cause issues with lateral impacts, so depending on the amount of uneven wear expected between wheels, it would probably be better to just assure that all of the wheels are in the same condition and the axle is attached to the frame soundly.

[SpeedFreed]

The steps I always take to make sure all wheels touch the ground are:
1) I don't make perfectly rigid drivetrains (on purpose!). You shouldn't be able to notice any flex, but it should be there in ever so slight amounts.
2) Mecanum wheels that flex, such as Vex Pro Mecanum wheels, actually have a small amount of give, so they could be ever so slightly off.
3) Diligence. I always use a CNC machine or mill with DRO to drill holes in my mecanum drives to make sure they are exactly in the right spot I do this for many other parts of the robot too).

But I can't imagine it would be all that different from a typical 4 wheel mecanum, just doubled.

[pimathbrainiac]

Interesting idea, taking off my/cjl2625's concept. How would the suspension work on this, since each corner of mecanum wheels are joined by bevel gears?

[SpeedFreed]

Well on first thought, you could have to pivot one mecanum wheel about axis of the other. While you wouldn't want to do this for anything extensive (I can't imagine mecanum wheels work well on significant angles) you should get a safe 1/8" of travel out of it, which would be plenty to account for any build tolerances. Then you just have to make sure the 4 solid mounted mecanum wheels are mounted such that they are coplanar (like any mecanum drive).

But, in my opinion, suspension is unnecessary complexity that should be avoided whenever possible. One of the goals in this idea was it's lack of articulation for simplicity and weight (and time and money...).

[Caleb Sykes]

Interesting, now I want someone to build it to test how it performs compared to a standard mecanum drive.

Unfortunately though, I don't think that the added weight/design complexity would be very worthwhile. If I am understanding this design correctly, any gains you get in the strafing direction will come at the expense of the forward backward direction.

For example, under a normal 4-wheel mecanum setup, top speeds might be something like 10fps in the forward direction and 5fps in the strafing direction. With the same reduction in your setup, my guess would be that you would see more along the lines of 7fps in both the forward and strafing directions (whichever directions you so choose those to be).

I suppose that if you wanted the robot to handle more like a swerve drive than standard mecanum, this setup could help. I know our driver did not like field-centric code with mecanum wheels since his max velocity was dependent on which way the robot was facing.

[Toa Circuit]

Correct me if I'm wrong, but wouldn't you be better off going to a 4 omni setup? You wouldn't have the losses from miter/bevel gears, it would be lighter, more robust (it seems omnis stand up better than mecanums), have the same strafe power as forwards-backwards, wouldn't require a suspension, and would be mechanically simpler? It's effectively what you're doing here already.

[SpeedFreed]

It is true that basically this only improves the strafing abilities and adds symmetry. While I can imagine it must have some effect on top forward speed, exactly how much is a question that can only be answered in testing. Also, speed can always we compensated for, like my teams new octanum--powered by 6.66 CIMs and does 22fps on mecanum (after mecanum loss). Likewise if you power each corner of this with 1 CIM and 1 Mini CIM, forward loss will probably be quickly forgotten.

As for 4 omni wheels, I'm not quite sure what you are getting at. If you mean replace one set of mecanum wheels with omni wheels, then you would need at least 6 independent gearboxes to drive all the wheels (4 for mecanum, 2 for omni wheels) (or differentials, but THAT truly gets in the realm of unnecessary complexity), and then that is some strange mix of H drive that really holds no advantage. If you mean just 4 omni wheels, then that would have to be holnomic drive, which only reaches top speeds on the diagonals (unless you do 4 omnis Killer Bees style, and then you have to be magic to drive it).

The biggest pro's I was getting at with this, is that it provides full motor power in the cardinal directions (unlike H drive), and is symmetrical like swerve. But it is also heavier, more expensive, and harder to make than mecanum, for probably not enough gain.

However if your team does mecanum every year without exception, and is tired of it's asymmetric performance, then this is the drivetrain for you!

[jman4747]

Originally Posted by SpeedFreed

As for 4 omni wheels, I'm not quite sure what you are getting at.

Originally Posted by Toa Circuit

Correct me if I'm wrong, but wouldn't you be better off going to a 4 omni setup?

Did you mean 2 sets of 4 or 8 total? Like this:

[Certified Nerd]

Why not just use 8 sets of 2 omni wheels? Wouldn't this work fine and be much cheaper?

[jman4747]

Originally Posted by Certified Nerd

Why not just use 8 sets of 2 omni wheels? Wouldn't this work fine and be much cheaper?

Do you mean 2 sets of 4 = 8?

8 wheels*2=16

16*$18 a wheel (4" VEX Pro omni) = $288

2 sets of 4" VEX Pro Mecanum = 8*$30 = $240

8 total 4" VEX Pro omni wheels = $144

[JohnGilb]

I don't think this really helps you.

So, here's the thing (Ether or one of the mecanum experts can correct me here): in an ideal environment, there is no difference in power applied forward/backward vs strafe left/right. The reason you see bias in the real world is because of the internal friction in the roller bearings - this effectively means the wheels aren't providing power at 45 degrees, but rather some other number that is typically biased towards front/back.

All you're doing here is combining two imperfect drives orthogonally, so instead of max speed in one direction and 60% in another, you're probably going to have 80% in both, as the fast drive will be "dragging" the slow drive.

[Lil' Lavery]

Draw out a free body diagram of both a typical mecanum set-up and then this one.

[lovelylily06]

What size mechanum wheels were you considering for this design? I think the efficiency of the design would be changed by the robot itself and not just the theory of the drive system alone. I generally consider the drive system more of the support but I really like this design!