6 wheel holonomic drive feasibility

[Ether]

Originally Posted by ekapalka

for all intensive purposes, the mecanum wheels could be plaction wheels

What do you mean by "intensive purposes" in this context?

[ekapalka]

I just meant that no matter how much power or traction the robot has, adding two more wheels and 2-4 more motors will add at least some additional traction and/or power (at least in our application). I'm fairly certain that that's what I meant :) We have no plans to go back to plaction (and I don't think it would work too well).

If it's the messed up figure of speech you were asking about, I would like to accredit that to auto correct (it's so intrusive on Android phones). For all intents and purposes.

[evanperryg]

Originally Posted by ekapalka

How about this: a 6-8 CIM drivetrain (not necessarily FRC legal) made for pushing (i.e just throw on some extra motors for forward pushing). That HAS to add at least some additional pushing/pulling power, regardless of how much traction mecanum wheels have (for all intensive purposes, the mecanum wheels could be plaction wheels).

The thing about mecanums and omnis is that they are designed to slide. That, in fact, is the defining aspect that makes them different from any other wheel. That extra omni might help you push forward and backward(emphasis on 'might'), but it will be absolutely useless pushing sideways because the rollers are passive. If you want to have omnidirectional drive that can push, you have 2 options: swerve or octocanum. And, in order for either of these options to be effective at pushing, you have to execute the design flawlessly. You are right, more motors adds more pushing power(if you do it right) but most of that power will be negated by the fact that the wheels are designed to slide around.

Originally Posted by T^2

You're completely wrong. Not only are there significant mechanical downsides (weight, size, complexity, number of gearboxes required) to mecanum, but they also come with a huge opportunity cost. The time and money you waste on those things could be much better spent training your students to make a drivetrain that works, and what's left over can be spent on the rest of your robot.

I feel like I remember seeing Karthik posting something that was very similar to this... Anyway, I agree fullheartedly. In addition to what you mentioned here, they also can have some severe mechanical issues. We did mecanum once, back in 2010 (first and last time doing holonomic). Somehow, something got bent so we could only strafe. That was also the heaviest, the least reliable , and the most complicated drivetrain we have ever built.

In my opinion, if you want to do a holonomic drive, mecanum is possibly the worst option. The wheels themselves are expensive, it offers no potential for pushing, and it is a nightmare to program. A well-executed tank drive and a decent driver will maneuver just as well as a typical mecanum drivetrain. If you really, desperately want to do holonomic drive for some reason, I would suggest either swerve or octocanum. Remember, both of these are extremely complicated, weigh more than tank, and without a good driver they will not make your robot any better.

How many bots that won champs/IRI used mecanum? None.
How many bots were even on Einstein that used mecanum? None.
How many mecanum bots were invited to IRI? One, and it was probably the best executed mecanum drive this year.

[Ether]

Originally Posted by evanperryg

[mecanum] is a nightmare to program

Could you elaborate a bit on that? What did you have in mind?

[ekapalka]

Originally Posted by evanperryg

[...] and it is a nightmare to program.

The only thing that makes me feel any better about programming mecanum wheels is this information (courtesy of Ether; thanks, by the way) and the mecanum classes in WPI (we're using C++), which I have been reading over and over since I found out we had the mecanum wheels (that doesn't mean I should rely on it, but it gives us hope that we can program it). We already have four wheels, four gearboxes, four motors, and four Talon SRs, all with no dedication at the moment. We're currently designing the chassis with room for a central gearbox (see original post), not necessarily for omni wheels, but for the possibility that we go back to skid steering. Hopefully it'll be water or plasma cut from sheet metal, so adding in additional holes for such things won't be that much more difficult (if we're getting it water-jet. If it's plasma cut it won't quite be as easy). Thanks for your input, by the way :)

[Oblarg]

Mecanum is not a nightmare to program. It has been done many times before; the knowledge is there, and so is the code itself if you wish to download it. It becomes somewhat harder if you want field-oriented control, but the difficulty there is with the gyro, not with the actual drive.

Mecanum is a fairly good, cheap solution to the omnidirectional problem. It is not a 100% solution; it has notable shortcomings (effectively lower coefficient of friction than traction wheels with the same material, lots of frictional losses when moving sideways, losses due to imperfections in wheel design/construction), but also notable benefits (ease of construction, ease of maintenance, relatively low cost). It is not going to compete with a swerve drive or with a hybrid drive, but it will give you an agile robot (acceleration in FRC, with the exception of 2009's silly teflon wheels/driving surface, has never been traction-limited) and that can strafe. For some games, this is all that is needed; 449's use of mecanum in 2008 was a huge asset for our team, as the anti-roadblock rule in that game made the lack of pushing power a near non-issue. For others, it is not.