Swerve implementation for teams doing swerve for the first time

I’ve seen a large amount of teams sending it with swerve for the first time this year (presumably due to the increased amount of time to work on things), and I have some questions for everyone who is making the jump.

  1. What made you decide to do swerve now?
  2. What module(s) have you used?
  3. What problems have you run into (both with the modules, and user error) while implementing the swerve?
  4. How much test and drive time have you had with swerve now that you’ve used it?
  5. Do you feel comfortable enough running swerve in a regular season now that you have tried it?
13 Likes
  1. We have been talking about making the jump for forever but believe that using it was gonna make us really competitive for the challenges
  2. We went with the SDS MK3 modules
  3. With our custom frame we put some crossbars too close to the module so maintenance is a bit of a hassle but that’s our bad lol
  4. We’ve been driving on them for about 2 1/2 months now and have probably changed the nitrile about 3 times in the past 3 weeks so they’ve been tested quite a bit
  5. We are planning on using it throughout the regular 2022 season unless the game demands otherwise like 2016. We feel like we have given drivers lots of time to drive the swerve and get much more comfortable with it for next year
4 Likes
  1. We had some time and some use it or lose it grant money. And we’d seen swerve doing interesting things.
  2. Thriftybot.
  3. None so far…we’ve just built one unit to try them out and have more on back order.
  4. Zero. Not running yet. We’ve been working on other things once we built the trial unit.
  5. Until the nature of the game is known you can’t be comfortable with any possible mechanism. We’ll be training in an entire new drive team and will have them work on swerve and conventional tank once we get the drive, and the drive team, up and running.

TW

3 Likes
  1. We probably would have done it this offseason anyway since we had been considering it for a few years, but the extra offseason time and not needing to prototype new game object manipulators made it that much easier.
  2. A custom module. Being an international team with a CNC router, it’s a lot easier and cheaper for us to make the module custom than buying a kit from the US and having it shipped.
  3. We had some struggles getting angular and velocity control loops tuned and getting field-relative driving working. But overall I’ve been pleasantly surprised by how smooth the process has gone considering it’s our first attempt.
  4. We built four modules and put them on a test chassis for programming. The programmers got about a month with that and had it working pretty well. The practice chassis was then disassembled and the modules were moved to our 2021 robot, with which the programmers and drivers had a combined 3 days before our first offseason competition. They weren’t able to get field-relative driving working, but the swerve worked very well and doled out some serious defense in elims.
  5. At this point, I’d still say no. We’ve had a very good start, but I don’t know that I’d be comfortable trusting our season to it yet. (I expect some of my students might disagree with me on that.) But if we get a good amount of driver practice and programming time with it before next season we might be ready if the game calls for it.
3 Likes
  1. We had money and time to use on swerve, a competition season where on field failure had no impact and about which we didn’t particularly care, FOMO, and of course the kids were excited to try something new.

  2. SDS MK2 for the first test chassis, SDS MK3 for the 2021 bot.

  3. Kids had module ordering set up wrong in code for a while, this produced some unexpected results when they started auto testing. Oops.

  4. Probably 40-60 total hours of drive and testing time. We’re still pretty limited on shop hours.

  5. Yes, we’d run swerve in season if it’s feasible. Skid steer drivetrain inside the swerve drivetrain too spicy for us.

3 Likes

Moving back to 2018 for the first time I was involved in fielding a swerve in competition.

  1. Well, I got told I was helping 323. And John Novak… so uh?
  2. Custom modules, probably one of the last few teams using slip ring based modules.
  3. The biggest issue was that the only software mentor was 1000 miles away from the hardware :stuck_out_tongue: This meant when our coordinate systems were flipped between the gyro and the modules it was difficult to diagnose.
  4. I mean, one huge advantage is we have mentors with some of the most swerve experience in FRC.
  5. NA - we did.

The best advice I have for folks making (or looking to make) the jump is to draw out coordinate systems and understand what your sensors should and are reading in given states.

2 Likes
  1. Its @nuclearnerd’s fault. He “encouraged” us to do it.
  2. Modified Brendan Simon’s 5406 module.
  3. Addressed mods to his design in our final version. Most were minor.
  4. Nov. 2020-Present.
  5. We are comfortable now to use it in any future regular season, if it suits us.
6 Likes

Thank you for the replies so far, everyone! Keep them coming!

Some background on why I made this thread: I’m really interested in seeing if swerve is actually becoming more accessible and ubiquitous for teams or if a lot of people are making the jump but finding that it’s still more trouble than it’s worth. Of course the outcome of this CD thread isn’t to help me or anyone else make a decision on whether it is appropriate for their team to use swerve (my team did it once in the 2017 offseason, and while the swerve did in fact work, the wiring on the robot did not), but instead to satisfy my curiosity after observing the exponential rise in teams fielding swerve robots.

With more COTS swerve modules being available, it’s clear that a sizeable portion of the swerve complexity (the manufacturing) is greatly negated, but I’m curious if the remaining resources required to reliably run swerve are still a significant enough factor to make it not worth, or if people are finding that with one large hurdle removed, it actually ends up in a much more reasonable range of accessibility.

  1. Seeing the robowranglers swerving at Plano last year and wanting to do something cool after the cancellation of the season. And it cooks
  2. SDS Mk3 with falcons and motor spacers
  3. Programming took awhile due to new programmers + non-trivial programming. No problems with the mechanical design of our drivetrain (we tried rivets and were thoroughly surprised at how well they are working). We did have a problem with the swerve going over the edge of our carpet and the stringy bits getting stuck in the bevel gears. Had to take the modules apart a couple times and now we have duct taped the edge of our carpets.
  4. between 20 and 30 hours
  5. Yes. Building an offseason practice bot and then developing guidelines to be followed when implementing swerve in future seasons has made us very confident in it.
5 Likes

We wrote a white paper on this topic for 2020 as that was our first year using swerve.
ChimiSwerve White Paper
Our programmers wrote this out hoping it may help teams with issues we had and things they had to figure out along the way.

We used the SDS MK2 swerve modules and used them again for 2021 since we already had a set on our practice bot.
In 2020, before competing our driver put in a LOT of practice time, but took to it really fast.

We used swerve again for our 2021 robot since we really liked it and even if the challenges had not been what they were and we had in person comps, we absolutely still would have done it again. In 2020, we did get to compete once with it thankfully.

4 Likes

[quote=“Andrew_L, post:1, topic:393143”]

  • What made you decide to do swerve now?
    We started swerve the summer between 19 and 20. We had a few team members that wanted to take the challenge on.
  • What module(s) have you used?
    We researched 2767, 33, MK2 and a few more. We designed our own with the help of our main sponsors inhouse manufacturing capability.
  • What problems have you run into (both with the modules, and user error) while implementing the swerve?
    We had issues with encoders and bevel and ring gears. The programming consumed a lot of time, I believe there are many more code options available this year.
  • How much test and drive time have you had with swerve now that you’ve used it?
    Prior to our first event we had less than 25 hours drive time. We had a team meeting prior to our first (only) event last year and the team decided to use our swerve. We had gear and encoder issues that caused auto issues and drivability issues, this hurt our ranking during the event, but also created the desire for continuous improvement by the team members. We now many many hours of testing and driving hours on our swerve modules.
  • Do you feel comfortable enough running swerve in a regular season now that you have tried it?
    This year we improved the yoke, gears and wheels and are very happy with the solution. We still have a minor issue with encoders, but have a working solution and a plan forward. We are very pleased with the swerve project outcome and will continue to move forward with development / improvement. It has been a great team engineering learning process. Given usability from a field perspective we will be utilizing swerve in the future.
    [/quote] .
2 Likes

* What made you decide to do swerve now?

Most of the reasons not to do swerve are gone or (likely) going away in 2022.

  • PDP expands to 20 slots, hopefully rules allow for 20 motors, giving up half your motors to drive was a hard choice, giving up 8 of 19/20 is a lot easier.
  • COTS swerve appears mature with multiple vendors selling modules and many of them doing so for multiple seasons with many teams running them.
  • WPILib Swerve software support and other teams sample code has lowered the barrier to entry to programming for swerve
  • No Bag(!!!) means more time to develop during the season in future years and get lots of practice on the actual competition robot so more complex controls and skills can be learned.
  • Hopefully the BOM stays gone, meaning you have to jump through fewer hoops to get a legal COTS swerve robot.

This season being different makes it a great year to experiment without having to recover mid-season if it doesn’t work.

* What module(s) have you used?

  • We haven’t built a robot yet but we have SDS MK3’s in hand, running Falcons for drive and steering.

* What problems have you run into (both with the modules, and user error) while implementing the swerve?

  • Haven’t actually built it yet, but it has been interesting designing around it
  • Like the intake can be on the “side” of the robot
  • corners of the the robot are a little trickier to build up than normal

* How much test and drive time have you had with swerve now that you’ve used it?

  • None yet but we should have it competing at The Texas Cup in June, and then we plan to do the skills challenge auto runs with it over the summer too.

* Do you feel comfortable enough running swerve in a regular season now that you have tried it?

  • The goal is for Spectrum to be ready to use it for 2022 if the field is reasonable (not 2016 basically).
9 Likes

You have no idea how many times I had to remind Marshall that “The enemy gate is down”

Swerves have no true front, it opens a world of new possibilities for multi modal driving interfaces where your front changes depending on what game function you are doing. Imho it’s one of the least utilized benefits of swerve.

14 Likes

We are doing the “summer of swerve”!

  1. What made you decide to do swerve now?
    After seeing lots of swerve teams doing well at the Texas Cup and our lead programmer’s constant “I wanna swerve!” it was time to consider it. Its one of those things that really can’t be done while the bot is being designed and built; its too much time and risk to commit to. Thus, if we were -ever- going to swerve, we needed to do it off-line! Plus, lots of commercial swerves out there now!
  2. What module(s) have you used?
    The “2021 winter swerve” from Nick 2021 Winter Swerve A key reason the kids chose this was that we could 3D print a large fraction of it and get driving. If we end up liking it, we can upgrade hardware and printed parts. I’ve got a pile of Carbon fiber filled ABS that I’ve been dying to try out… And now I’m printing with it!
  3. What problems have you run into (both with the modules, and user error) while implementing the swerve?
    Still working toward having hardware. Bits are showing up daily, and I’ve got most of the printed parts for the first corner.
  4. How much test and drive time have you had with swerve now that you’ve used it?
    None yet…
  5. Do you feel comfortable enough running swerve in a regular season now that you have tried it?
    TBD. Hoping the answer is “that’s good to go!”
  • What made you decide to do swerve now?

Commercial availability and the sample code that SDS provided. We’ve talked about it for a while so the pandemic was a good time to put up or shut up about it.

  • What module(s) have you used?

MK2 and MK3

  • What problems have you run into (both with the modules, and user error) while implementing the swerve?

Mechanically no issues. The modules actually made our overall chassis and drivetrain assembly way more simple and faster with less parts and pieces to manufacture. Coding wise, we still struggle at times so stuff like auto wheel alignment reading values off of sensors are things we struggled with. Initially, the amount of tred wear was concerning but it really turned into a non-issue. We changed tred once a day at Texas Cup.

  • How much test and drive time have you had with swerve now that you’ve used it?

A ton. We’re lucky to have a full practice field and we had a full in person offseason and build season.

  • Do you feel comfortable enough running swerve in a regular season now that you have tried it?

Yeah we took the robot to the Texas Cup and our swerve performed well. We have some things we want to tweak a bit but generally speaking the MK2 modules (what we used at competition) were plug and play with no problem.

1 Like