Swerve is a distraction

Technical Technical Discussion
1

Swerve is a distraction for most teams, right? Its sort of necessary if you’re going for big time in Houston. But for most teams its not at all necessary and pulls energy away from improving other mechanisms, auto, and sensor based controls.

Several of the students on my team are obsessed with swerve (we’ve always been tank). They seem to see it as the entrance to the big leagues. Good old consumer culture of buy your way to a new, better you. In reality our team has been a bottom third team. And what is most shocking is last year we came up 6 out of 44, with a tank bot! We had a great auto. We had a low CG for stable driving, lots of drive practice, great mobility (for tank), and also did great on the final balance. We only did cubes and did them OK. Essentially a simple bot that was done well - KISS. We got really really lucky but that’s a different story. The dominant lesson from 2023 was a simple bot, executed well can be a top ten regional team. There were a dozen or more swerve bots ranked below us.

I’m taking some heat but I really don’t see the value of us taking 3 or 4 of our strongest students and putting them on swerve. I think there are far more valuable skills our team needs to develop first.

12 Likes
2

An average tank drive team in this past 2023 year game was no match for a good swerve robot. The scoring grid with the charging station in front made this a no-brainer choice. There are always exceptions, but this sounds like a generalized OP.

27 Likes
3

image
image1125×719 214 KB

Swerve is not needed to be a competive robot at district/regional level. Swerve is not needed to be useful at a championship level.

20 Likes
4

The most basic requirement to play an FRC match is the ability to drive. If pursuing swerve prevents you from reliably driving every single match, then yes, it would be a distraction for your team.

40 Likes
5

There are a couple of reasons why not to do swerve:

  1. You don’t have the money
  2. You don’t have the programming capability
  3. You are a rookie team (or a team without a significant member turnaround from a prior year)
  4. It will inhibit your ability to drive your bot during competition (usually because of one of the things above)

Couple reasons to run swerve:

  1. It’s more accessible than ever
  2. You’ve done tank or any other type of drive in the past consistently
  3. You want to be competitive. (It’s a no brainer from a ‘what is more competitive’ point of view, although I’m not saying there aren’t competitive robots with other types of drivetrains.)

I think the main point is do what you feel like your team CAN DO RELIABLY. If you like swerve awesome research it and run it. If you feel like tank drive is the way to go for you also awesome run it, but the mindset that swerve is a distraction seems highly irrational and I would argue that everyone in FRC at least researches it! (bc its cool)

EDIT: Forgot to add that the main thing that will make your robot competitive in competition is focusing on your design and a FOCUS ON RELIABILITY.

19 Likes
6

Swerve is definitely an amazing capability, and I think no one denies that. Developing a swerve in a build season is a stupid idea in my opinion because you are so time constrained. Developing a swerve in the off season becomes a much more attractive choice to me. In the modern age, swerve is not that hard to pull off if you have the money to buy it, and can improve performance a lot. There are so many libraries that work, and little to no manufacturing has to happen in the shop. I think it is more maintainable in the year than any 6 wheel, having done many configurations, and swerve. Saying this, there should be no development time on the swerve in the season. If that can be a thing your team can do, I would consider it, but if you are going to spend time in season making it work, it isn’t the optimal decision in my opinion.

7 Likes
7

Prior to the 2022 season, your argument would be correct and the vast majority of people would agree with you. There were many well-known, Einstein-level teams that preached the benefits of tank drive religiously. Those that did do swerve either had years of experience testing and refining designs, or would struggle to move all season. A well-made custom WCD or other tank drive was by far the best bang for your buck in terms of effort to performance ratio.

That paradigm has changed massively in the past two seasons, and will likely continue to swing in favor of swerve drive more and more, barring drastic changes in the game rules or other meta developments. The reason for this is well-made, low-effort COTS swerve modules. SDS, WCP, and Rev all have COTS modules available that take away all the difficulty of designing a swerve module, 95% of assembly, and at least 50% if not more of programming. It no longer takes an entire sub team dedicated just to your drivetrain . Almost any mechanical student (even first year students with no experience) can follow the directions to assemble the modules and have a rolling chassis in the first week of build season. Using one of the many sources of example swerve drive code, one or two dedicated students and maybe a mentor can have it driving by the end of week two, and you can start driver practice. It comes down more to cost than anything now.

That said, swerve drive may still not be right for your team. It is drastically easier now than it was just a few years ago, but there is still some effort required. I would highly recommend any team interested in switching to swerve build and program a test chassis in the off season, so they don’t need to expend that effort during build. The other factor is funds. The way I see it, if you goal is to maximize competitiveness, you should go with swerve if you can afford the cost of modules and motors, or the AM14U kit chassis if not. Anything else is putting time, effort, and money into what is essentially a solved problem. Building a custom tank drive is educational, sure, but at this point a swerve drive is achievable enough that if you lack the funds to go swerve, I would skip the custom drivetrain for a year and use the kitbot, save the funds, and go in on a COTS swerve drive the next year.

As a point of reference, you might be interested in this thread from the last season: Drivetrain Type Data
Swerve didn’t quite make the top of the list this year, falling just a few percent below tank, but I fully expect that come next season the numbers will shift to swerve being a plurality if not a majority of robots that see the field. However, if you look at numbers this year for elims, DCMP, and Champs, the scales clearly tip in favor of swerve. It’s more than a little difficult to separate causality from correlation here, but there is clearly a link in the data between swerve drives and competetiveness.

tl;dr:

  • swerve is easier than you might think, if you have the funds for COTS
    • if you don’t, use the kit chassis and save up funds for next year
  • putting a swerve drive together in the off-season is a very useful prelude to using one in season, but there are many examples from this past season of teams building their first swerve in-season without issue
  • COTS swerve can absolutely align with the KISS principle, and can have you driving a chassis week one
36 Likes
8

I’m afraid I’m going to be picking apart your reasoning. I don’t necessarily disagree with your conclusion, but your reasoning is… a bit flawed, shall we say?

  1. L.A. Regional has a surprising number of single-event teams–and a lot of the 2-event teams use it as their first event. It was your second event. You had a full event to learn how your robot handled and get good at driving. Some of those other teams barely had it working on arrival, let alone getting practice. Automatic performance boost.
  2. L.A. Regional had a fairly high proportion of tank drives–if you take a look at the link @Zaque posted, 14 swerve, 4 mecanum, 28 tank. Again, quite a few single-event and first-event teams, many of whom do build tank for various reasons. (Like no budget.)
  3. I kinda hate to pull this on you: You got lucky. OPR has you as # 9. (TBA regional page, insights tab). I don’t want to say “favorable schedule” because I can’t really pull those stats quickly.

In other words: you did that well by being better than most other tank drives.

And there were 3 above you, none of whom picked you. Of the 14 swerve bots: 4 alliance captains (1, 4, 6, 7), 7 first-round picks (# 8 picked a tank as their first pick*), and the first 3 picks of Round 2. Every single swerve robot came off the board. Interesting. (Ranking, IMO, is a bit of a crud-shoot. Picking order is often “where it’s at” for “how is it really?”)

How about buying a better 6WD? They are on the market. Just sayin’, you ain’t gotta buy a swerve to buy improvements. Or even a mecanum (OK, now I’M going to get the hate–for this game it might have actually been a decent option). Kitbot honestly kinda needs upgrades to be more competitive, and it looks like that’s what you’re running if I’m not mistaken. How about buying an elevator kit? (Back in my day “elevator kit” meant “the raw materials you buy to build your custom elevator”.)

You could argue for a return to the “limited COTS” days if you’d like to. I don’t think you’ll get farther than being laughed at by 2/3 of CD. (Full disclosure: Some days I’d kinda like to go back to then. Then I wonder why.)

I’m going to play “one-up” on this. Your sights are too low. World Champion is where it’s at, and I can think of 2 different World Champion robots in 2 different years that could both be a Minimum Competitive Concept**. The key problem I have with this statement is that the robot needs to be as simple as possible, but NOT simpler. If swerve seems to be a required stepping stone, or required part of game play, and you build tank only, you done screwed up!

Suggested thought exercise: do you think that if you’d focused on cones instead of cubes, you would have done as well at scoring the game piece with your tank drive, or would you have needed additional features? (Not necessarily swerve.)

Now that I’ve kinda ripped up your logic a bit:

You’re not wrong that swerve is possibly a distraction. For now. The historical consensus is that if you’re wanting to do swerve (or, really, any non-tank drivetrain) that you want to do it in the offseason before the build season. I don’t think that’s changed much, but it’s gotten a LOT easier to get started, so much so (with the libraries and COTS modules) that it’s actually possible to do it for the first time in-season and not accidentally make that the only robot you get done.

And if it’s going to take 3-4 strong students on swerve, when there are more valuable skills to develop first, then yes, you may want to focus on those skills. (Given that your team has the resources to pursue those skills, and all that.)

BUT. With the way things are currently going, it’s likely to end up where teams without swerve are rookies and “perpetual rookies”***, who consistently get clobbered by swerve teams. At that point, when having swerve is a requirement just to be competitive at all, then teams that don’t invest the resources are going to be in a world of hurt. And I don’t think that day is far off–doubt it’ll be 2024, but by 2027 it’s going to be close, I think.

And when that happens, tank drive may end up like mecanum and tank tracks are now: Relatively rare, and shunned on appearance. (At least from non-rookie teams.)

One more tidbit: I can tell you that a minimum of 2 teams at L.A. who ran tank last year have swerve drives available for testing right now, with the intent to go all-in next season. One ran at an offseason, one didn’t quite finish.

*From a standpoint of “look at the rankings”–I’d guess that this is a “crud we’re in and we didn’t think we would be, pick the next team down and hope they’re good”. Happens a lot. Possibly even for the next pick that alliance made.
**148 in 2009: 3-wheel swerve drive with a pole. IIRC, # 2 pick of their alliance. 330 in 2005: single-joint arm on a 6WD 1-speed drivetrain, with no power past the pivot of the arm. Alliance Captain. Minimum Competitive Concept: Bare minimum robot to reliably make a decently-deep run in regional elims.
***Come find me next time we’re at the same event, say your first event next season. I can tell tales…

20 Likes
9

I can!

6658 was 15th by Expected Points Added (EPA), which I believe to be a slighter better measure of individual robot performance than OPR. You can read about the differences between OPR and EPA here.

Additionally, 6658 had a Strength of Schedule (SOS) score of 0.44. 0.50 is the mean schedule, so while on the easier side, the schedule wasn’t wildly easy. This was the 16th easiest schedule at the event, which would put them in the 36th percentile by difficulty.

My $0.02 on the matter at hand - teams are made better by scoping their robot design to be within the bounds of their resources and abilities. Likely, swerve isn’t outside of your abilities, but it may currently be outside of your resources. I think you’re doing a great thing by focusing simple – that’s absolutely the right track to winning and improving. I don’t think that swerve is “a distraction”, so much as it should be treated as the next stepping stone towards being competitive. It’s not the boogeyman it used to be in terms of implementation. If you can scrounge up the funds in June, it’s likely to be a very worthwhile investment to test with until January.

18 Likes
10

Drivetrains don’t score (most of the) points in matches. Mechanisms score points in matches. Are you building high-performing and reliable mechanisms that score lots of points in matches? If so, you can improve your performance with swerve. If you’re not, then put your effort into making high-performing and reliable mechanisms that score lots of points in matches and stick with a standard tank drive. A fancier drivetrain won’t allow you to score points better, and with the economic law of opportunity cost, any resources invested into swerve will be resources that ultimately cannot be invested into making high-performing and reliable mechanisms that score lots of points in matches.

Build high-performing and reliable mechanisms that score lots of points in matches on a tank drive first (and execute on them in season), then make the switch to swerve. If you jump to swerve before this, you’re going to score fewer points in matches, which can convert a lot of potential wins into losses.

6 Likes
11

For many games, it’s a lot easier to make a good mechanism if your drivetrain is holonomic. Games like 2017-2019 come to mind, where robots with swerves could get away with worse mechanisms for the same result.

Nowadays, it’s so achievable to make a swerve drive, I think any mid tier team will find it easily worth it if they can drop thousands of dollars on modules. I’ve also found that drivers train much more quickly on field centric swerves where they don’t need to care about direction or be particularly careful with their turns.

14 Likes
12

Just want to reinforce what @asid61 said. If your drivetrain has an extra degree of freedom, that oftentimes will eliminate a required degree of freedom from your mechanism. Many teams this year used their drivetrain to line up with the cone nodes on the grid much faster than if they didn’t have swerve, and assuming a COTS module at a much lower overhead than if they had to add an extra degree of freedom to their mechanism.

6 Likes
13

I want to also add to this. You still have to build a good robot even if you make a “low goal” (that’s what we call them, this year it’d be a cube bot) robot. The “simple” robots that a lot of people reference here are still very hard to reach quality wise for a majority of teams. I feel like it’s an unrealistic expectation to just slap together a robot that does a task really well and pretend that it doesn’t come with any effort (this isn’t directed at asid).

To win you need to a consistent, robust, and well driven robot. You need a good auto and game strategy to break top 8. This isn’t to say that you won’t have a better winning record to previous seasons but to say that a team will be more competitive by being a low goal bot is a bit of false hope. To go further, I have seen a miss match between this advice and actual results of regionals. I do have some stats but naming specific teams feels a bit overly critical for the point. But a few declines came from low goal bots ranking high but other teams not wanting them. This has an affect of that teams having a better winning record but at the same time being declined in alliance selections which feels bad. On the flip side, the opposite has happed with quite a bit of frequency. 2910 in 2019 comes to mind.

tl;dr

Results vary. If you are low resource team and want to be competitive build an everybot and practice driving religiously. An everybot is always nice to have in most situations.

3 Likes
14

With the 2024 season upcoming teams will have an option of purchasing a serve from whichever vendor they want and if they are using a CTRE motor + sensor system; running the Swerve setup tool and then barring and misassembly of their mechanical components have a running and driving swerve in under two hours.

It’s no longer any more difficult than a kit bot drive.

14 Likes
15

Sounds like this is the real main issue - your team does not have consensus on what’s most important to work on.

10 Likes
16

This sort of stuff is really unhelpful. Why can’t we let teams set achievable goals for themselves instead of dunking on anything less than “world champion”?

A lot of what you said is very useful and fair advice - swerve should increasingly be in team’s roadmaps for achieving their goals. I’m objecting to your phrasing, which does matter.

18 Likes
17

Swerve definitely takes resources to pursue including a fair bit of cost. However, I am of the opinion the time required for setup is extremely worth the payoff. It is absolutely possible to be a competitive robot with tank drive (see 3357 this year). However, if you have the time to learn to set it up and get some practice drive time, I think it is one of the single most beneficial “mechanisms” you can practice to improve your team’s success.

I also think that teams vastly exaggerate the time it takes to setup your first swerve drive. There are so many resources available for learning how to setup your first swerve drive.

1 Like
18

Hi Eric. My team is one where this question was asked annually until we built our first swerve robot, in the period between the cancelation of the 2020 game and the reopening of 2022. We still used a custom WCD for our 2022 drive base and like you I felt we were not hindered by it except at the highest level. We built another swerve in the fall before the 2023 game and ended up using nearly the same design in season.

Currently my team is assisting a rookie team and a low resource team in their off season work, and we gave each of them swerve modules (REV and SDS) to build with. Because of the quality of both the currently available components and the plug-and-play software, two of the three biggest hazards in doing swerve are no longer a problem that should prevent a team from using it. And it is in fact easier to design a drive base, easier to construct, and easier and more intuitive to drive. The remaining obstacles for any team are money, and, working on it (or any new system) in the fall.

Since you’re asking this question in the fall, it sounds like the only hindrance you still have might be money. Glad you’ve asked the question! Good luck to your team this year.

13 Likes
19

*I think swerve should be a top priority for basically all teams at this point. If it’s not viable this year for any reason, you should be identifying that reason and looking to address it. The path to being competitive with a tank drive is so narrow, especially for teams that are on the edge of getting picked at their local events. When it comes down to alliance selection swerve is (for many teams) very often a deciding factor between two teams on a pick list.

*This is assuming a team’s goal is to be competitive (which is the case for most teams that I know)

24 Likes
20

Games differ enough from year to year that taking the dominant lesson from one year and applying it blindly to the next year often doesn’t work well.

Simple fast low goal scorers were all the rage in Houston in 2023. They were not in 2022.

It would be hard to build a simple fixed-shooter (no turret) tank drive bot that performed as well in 2022 as a simple low goal scoring tank drive bot performed in 2023, especially considering that in 2023 a tank drive could be 100% effective in the end-game, but in 2022 you had to build a separate climber.

2023 was just a really well designed game when it came to including a place for simple robots.

17 Likes