Concerning Proliferation of COTS Swerve

Did you stop by the pits of the other teams with COTS swerve modules at your competition and ask their team members about their experience getting the system to work, before assuming that their reasons for obtaining a COTS swerve must have not been very good on the whole?


While the material cost is low, the labour time was quite high. My team is very fortunate to have a balance between the relative freedom of a community team and just enough budget to fund the project of the a few years. For the argument of to buy swerve if a team isn’t able to make their own due to time, location, machine or other restrictions I don’t have a real answer. My gut feeling is that it should be earned through the work, learning along the way, but nor can I fault a team for buying COTS purely for a competative boost as it’s within game rules and a pretty logical move.


Another consideration: SDS are up to revision 4 (at least) of their module. They have far more hours of testing than any one team’s custom solution, and they’re still discovering issues. In addition to saving precious time, buying COTS gives teams the benefit of all of that OpEx.


I’m holding out for a SaaS offering, Swerve as a Service.


Subscription based, cloud hosted, and AI powered!


So… Pay a subscription and get monthly “updates” that break all the functionality? Like you wake up one morning and your steering motor is gone because they “fixed” a bug?

Like Toyota owners who found out their remote start feature was going away?


I did speak to some (not all) of the teams, nor do I assume their reasons or intentions immoral. No doubt it was a sensible, considered choice that they probably don’t regret making. My concern is for the direction of things, not only is a significant game challenge completely solved with a dense wallet with [relatively] less work and involvement, but many teams with the resources to purchase COTS swerve are also the ones definately with the resources to develop it. This development process is probably the single project I learnt the most from and most proud of, even knowing that many have achieved it before. My disappointment that many of the teams could practically have gone down the route of development but chose not too, in my opinion to the detriment of fellow student’s experience and the FIRST program’s goals.


Or the BMW owners that had their heated seat subscription expire?

I see more spam to rival car warranties on the way.


It was interesting how many teams opted for inverted modules that I saw this last regional, which is an iteration in itself.

My biggest hope for swerve is that it becomes more affordable or at least secondary market gets large enough, that lower resourced teams aren’t locked out when it is a competitive advantage.

Making custom swerve is still open to teams, and may be a way that it is more open to teams. Especially if easy to manufacture, low-cost options are made available, with better step-by-step builds. Typically however fully designing a module does take a number of design iterations and testing to be competitive, and that process while good for students, has mostly been too intensive or costly for the majority of teams.

I was very interested in custom swerve drives during 2019, and visited some of the innovative and best performing teams at champs, and learned about some of their processes, including having a student talk about the 50 iterations in CAD, or 3-4 years of testing before using in competition. I sort of knew this but it was eye opening about how difficult this was, and also now how much COTS has progressed this capability to many more teams that wouldn’t otherwise be using it.

1 Like

This. :point_up:

My team resisted going to swerve drive even after several seasons playing with and against our good friends 2767. About five years ago we started experimenting with our own swerve module designs as off-season projects. When those got to the prototype stage, we took a hard look at our own design vs. SDS, and decided to go with MK2. @NoshBrooks can say much more about that thought process, but the short summary is that we wanted the benefit Brendan talks about above. Our 2020 robot was the first in our ten year history to feature swerve drive. Having that and a year of continuous improvement (to put a brave face on 2021) behind us made it easy to select swerve drive for Rapid React.


Pardon me sir, do you have a minute to talk about your swerve warranty expiring?


Yeah, we seriously thought about just asking 2767 to build us an extra copy of their drive train. Something, something, fabrication rules.


A lot of different people want FRC to be different things. I’ve always been a big believer that the proliferation of COTS parts is a good thing for the program. Same goes for design sharing projects like Ri3D and the Open Alliance. However, there are folks out there who think FRC should be more about designing from first principles; severely limiting COTS options and making design sharing taboo. The idea is that teams should all start from the same spot and succeed on their own resources and abilities. Unfortunately these two schools of thought are dissonant from each other. It’s hard to find a middle ground.

I’ve never been certain as to what FRC should be. Sure I have my own idea of what serves the program best, but I’m willing to concede that I might be totally off base. Maybe there’s a world where in FRC you just receive a box of parts and a list of some other parts you can use, and specific allowances on how you can manufacture. Then you build in secret. Heck, some people have suggested you receive these parts at the competition and you have build on site. Some even go so far to say that mentors shouldn’t be allowed to help, making it a true test of student ability. Definitely not what I think is best, but I’d sure want to watch.

So while I don’t agree with the OPs perspective, I think there’s still a lot to be gleaned from what they’re saying. And I get why someone who with those viewpoints would be frustrated by the ongoing COTS revolution. It’s the same way there’d be outrage on this site if in 2023 there were severe restrictions on the usage of COTS parts.


I think a similar argument could be made about Limelight and vision processing. Yes it expensive and many high tier teams are now using it because of its optimizations and nice package. Many of those same teams also did or could’ve developed their own vision systems with OpenCV, but now get to use Limelight instead. It does take away some of the challenge of doing vision, but it also allows teams with not as good programming teams to also do vision processing. I think swerve is following a similar route, with high tier teams that have already developed swerve switching to COTS swerves because of the reliability and while also allowing lower tier teams do swerve.


Weren’t you a student during those days? Many robots were awful. Just driving was a major accomplishment for rookies. I may be a grey-haired curmudgeon but you won’t hear me advocate for going back to that world.


I am interested to hear your perspective on what FIRST should be, if you’d be willing to share.

Our swerve CAD is public on CD with a thread following our entire development, as I believe FIRST is significantly about gaining knowledge, experience and expeciences in engineering or similar feilds. From that perspective a COTS gearbox furthers this, though COTS swerve can but fairly often is to it’s detriment when the resources could have been put elsewhere for more significant gains. On the competative side of FRC, it becomes a paywall that many simply can’t get over and is very hard to engineer a competative alternative.


Small Parts for the… Win? I hated those days.


A nice balanced post. Thanks.

Within 2767 one thing that is hammered home to all students in all sub teams is we are in this together to make the best overall product. Never get married to your ideas when we are all working toward the best possible product. The robot is the “product” but so is the approach on how to compete with passion, I think there is a TED talk on that somewhere

This can be extended to the greater FRC community for the development of the FRC experience. The TED talk is probably worth the watch again.


Do not speak ill of the great rainforest in the sky.


I was, and it’s a big part of why I’m such a huge fan of the Kitbot, and why 1114 developed the Kitbot on steroids, and about a dozen other major projects while I’ve been a part of this program. FRC is really hard, and I firmly believe that anything that makes it easier is a good thing. We’re not going to access the communities that really need this program without lowering barriers to entry. That being said, not everyone has the same goals or desires for this program.