AM' swerve modules

[Tom Bottiglieri]

I'm all for it.

We should try to raise the floor, not lower the ceiling.

[Aren_Hill]

Originally Posted by Tom Bottiglieri

I'm all for it.

We should try to raise the floor, not lower the ceiling.

ugh time to pull out the 6wheel swerve actuated center drop cad files
didnt want to resort to this

[EricH]

Originally Posted by Aren_Hill

ugh time to pull out the 6wheel swerve actuated center drop cad files
didnt want to resort to this

I'll just grab that 6WD mecanum drive that's been posted... might make it rigid, though.

[IndySam]

Have you seen the new FP Planetary Gearbox?

[Alex.Norton]

I would say that something like this could have its advantages. The problem of making a specific swerve module does present itself, this is a little too specific to really sell for a first competition. However, I could easily see AM selling a generalized turret module because whenever I would thing of making a turret, or swerve this was always the area that I got hung up on.

Don't get me wrong, i'm never a fan of standardizing any piece of the robot. I always built a custom frame when I was a student on a team, this was one major reason that I was never a fan of the kit bot. I was always disappointed when I walked into a team pit to find that they had used the kit frame cause I didn't get to see anything new.

But to see the other side of the argument, I've been working on computer vision and specifically motion tracking on a mobile platform. The overall idea of the algorithms behind this task (ie. optical flow, image segmentation) are well within my grasp as a sophomore in college. However, I could never apply the linear algebra or the mechanics of the algorithms behind these, which is why I turn to libraries like Opencv which has the mechanics already finished for me. Yes I could learn a lot from writing my own mean shift segmentation on an image, but if I use the library I can write some very cool code much faster. In the same way, if AM offered a swerve module a team could get a working robot much faster and perhaps be able to implement some very cool control algorithms.

[GUI]

One thing that hasn't been mentioned is actually interfacing such a product with team designs. How would such a module be designed to be mounted on a team's frame, given the great variety in design and construction techniques between teams? I don't think it would be possible to make a module that fits the size, cost, and mounting constraints of enough teams to make it a commercially feasible venture.

[Tristan Lall]

Originally Posted by EricH

If AM were to sell a complete swerve module...

...Would it not fall under the following?

The thing with that paragraph (and also the one preceding it in the beginning of 8.3) is that it's unreasonably vague. I suspect FIRST puts it in there as an obstacle to those who might attempt to bring such a thing to market, and as a warning to teams about such products, rather than because they think that it's enforceable with any sort of consistency. (It's not even a numbered rule, which makes it doubly unclear how it's supposed to be enforced....)

When the test is effectively "I know it when I see it", the inspector doing the inspecting should be allowed to be flexible about the interpretation. Generally speaking, the most equitable way to enforce loosely- (or worse, poorly-) constructed rules is to give the teams the benefit of the doubt.

Plus, even if a COTS assembly was intended for use in FIRST competitions, and provided a complete solution for a major robot function, and was thus impermissible, it could still provide a function that could be used in many different ways, rendering it acceptable. There's no priority given to one condition or the other, so which is it: legal or illegal?

If the "rule" isn't a rule (in the usual sense), is blatantly unclear, and enforcing it requires a judgment of the designer's intent, I don't see any way that this can be reasonably and consistently applied over the range of possible COTS parts. That's different from being a safe bet, however: FIRST may think that there's no confusion, and act accordingly.

The economics of the situation look good, though: charge $250 per module, and sell some accessories that can be arranged in several different configurations (like different motors, different wheels, or mounting strategies), and you've got a winner.

[Mr. Freeman]

So, why aren't the andymark mechanum wheels against the "spirit of the competition" but a swerve module is? After all, by using AM mechanum wheels teams learn nothing about machining, CNC usage, roller molding, etc.
Same thing with the gearboxes.

(This isn't directed at anyone in particular, just a general question)

At some point you have to draw a line, what's the justification for putting it here? And what other things are prohibited by the same logic?

[MrForbes]

General answer: A set of mecanum wheels is not a "full mobility system". A set of swerve modules could be, or not, depending on how complete they are. A gearbox is not.

Using the examples in the rule is probably a reasonable way to evaluate it.

[Herodotus]

Originally Posted by squirrel

General answer: A set of mecanum wheels is not a "full mobility system". A set of swerve modules could be, or not, depending on how complete they are. A gearbox is not.

Using the examples in the rule is probably a reasonable way to evaluate it.

I can pretty well guarantee that a set of swerve modules isn't going to be moving anything without a frame attached to them and motors and gearboxes to spin them. I'd say under the current wording individual swerve modules would be permissible, especially if the wheels themselves are not included. (which I wouldn't expect them to be.)

As for rather it's in the spirit of FIRST I say it really just depends on the team. Some teams will have the resources to build a swerve drive and others will not. An Andy-Mark solution would not be the best choice for everyone anyways, so there would be plenty of teams building their own swerve drives. In fact I would hazard a guess that most teams that currently build swerve would keep building their own since they have it pretty much down pat.

Currently you can basically buy all of the components you need from AM to build a robot without doing any machining.(Wheels, gear boxes, and kit frame.) So I don't see the harm in adding a swerve module. In fact, I can see the argument being made that having such a thing available could be better for creativity and the design process. It adds another area of thinking into design. "Do I invest the time and effort into designing and build a subsystem to my exact specifications, or do I spend money, save time, and attempt to use an off the shelf product not specifically designed for my machine?"

Whether it would be economically feasible for AM to do it is a whole other story. Can't imagine swerve modules would be cheap to produce.

[Foster]

I'm all for AM to sell the swerve/crab/snake drive kits. I don't think that the basic kit would fall into the "COTS for FIRST only" there are lots of other roboteers that buy stuff from AM.

A few posters pointed out that the hard parts are not the swerve drive but the mounting onto the robot and the control system, and I agree. I don't see anything wrong with buying a part to make some robot assembly go faster. It's how its done in the "real world", COTS over custom, COTS wins a lot of the time.

When we first started out we traveled to Canada each summer to hand dig 50 tons of aluminum ore and 100 tons of coal. Our grant from Alcoa gave us access to their smelters. We turned our coal into electricity and used it to smelt our ore into a thousand pound aluminum billet. We then machined the billet into the parts we needed for the robot. With a small team this left little time to do detailed design or control systems.

Since then we've relied on suppliers like AM and McMaster. This has freed up lots of roboteer time to work on design, assembly and control systems. And it's paid off, our robot this year did not need any major repairs in the pits and we won an award for our design and control system.

We did look at swerve drives for this year and instead built our 6+1=3 drive that worked very well. But it would be nice to have a swerve drive option that we could bolt into place and spend the more important time of integrating it into the system.

[JesseK]

Let's think, for practicality's sake, about 'what IF' AndyMark DID make a generic swerve module. Let's think for just a second about the risk involved. AM put their shifting transmissions out, and for a time they sold well...that is, until 2009 rolled around. AM learned this year just what kind of impact niche products have on their ability to roll through inventory in FRC. How many teams bought Super Shifters this year? What if FIRST releases a game in the next couple of years that negates any need for a crab/swerve module in the future? The risks are great, whereas their efforts could be spent in other areas that FRC teams need. I for one am a fan of the hex bearing they recently put out ... such a small innovation that opens up a world of design for custom gearboxes built from COTS AM parts.

Besides, teams need to be able to create the partnerships necessary to manufacture something as complicated as a crab/swerve module. If you venture out, 30, 50 miles even just to find the right CNC'er or Waterjet facility then that's a relationship you'll use in other aspects of your bot -- not just the drive train. I say this because recently found a guy with a waterjet who's 20 miles away who we will negotiate a deal with for next year's bot. It took years to find him, but we finally did. Had we been given the items we needed for that particular robot piece this year, we wouldn't have even bothered looking for that relationship.

Finally, upon attending 118's crab module conference in Atlanta, then seeing the multitude of very different designs that were in Atlanta (111, 118, 88, 548, 1885, and several others), one would have realized that there is NO one-size-fits-all solution to crab design. There are too many design parameters, too many external systems effected, and on top of that many decisions have to be made for control that also effect how the crab module is designed in the first place.

[MrForbes]

There are a lot of ways to make frames too, yet AM and IFI each make a single frame kit that many teams use successfully.

The ability to design and fabricate a coaxial swerve drive is beyond most teams, and could probably be done by a company such as AM so that it's affordable, efficient, and somewhat universal. I expect that they would be open to suggestions for design parameters, hopefully from teams with experience with this stuff. It could be designed to be sort of universal, with a standard 6 hole sprocket hub for the steering, and a standard 1/2" keyed shaft for the drive, and allow the users to install their choice of standard sized 6" wheels with the standard 6 bolt sprocket pattern, or a common hex sized hole. The mounting flange could be designed to be easy to adapt to different types of chassis, including the kit chassis.

I've heard that Andy Baker spends a lot of time sitting around doing nothing, so this would be a great project to fill up a small portion of his free time.

[Jared Russell]

I have seen (very effective) swerve modules made with a single piece of 4" square tubing and a drill press. The assertion that you need a full machine shop to make a swerve drive is false. Yes, it might be prettier, lighter, and more efficient with the help of a CNC, but isn't that true of any mechanism in FIRST?

[EricH]

Originally Posted by Herodotus

I can pretty well guarantee that a set of swerve modules isn't going to be moving anything without a frame attached to them and motors and gearboxes to spin them. I'd say under the current wording individual swerve modules would be permissible, especially if the wheels themselves are not included. (which I wouldn't expect them to be.)

A few years ago, a team or few and a company were looking into pairing up to sell complete, disassembled tank tread drive modules to teams without a frame. This very issue was brought up, and as far as I know, the module still isn't commercially available to teams, even though the debate was never officially settled. Just sayin', there's different ways to look at "complete solution".

Now, if AM sold the parts for a swerve module, well, I can sure think of some uses for bevel gears or a swerve module's box section that aren't in the drivetrain, say, in an arm mount.