What are the upsides and downsides to having a swerve drive? I’ve recently been looking into making a swerve drive with NEOs, but I’m not exactly sure if they actually perform as well as what I think they can. I’m mainly worried about the cost, complexity with manufacturing/programming, and any issues with driving it.
Not to be rude, but there are dozens of threads from this summer about NEO swerve. Especially 2910’s. You should fall down the rabbit hole of the search results.
As a wise man once said:
Don’t be fooled by 2910, 2767, and 1323. They make swerve look like a beautiful, far superior drivebase.
For every 1323, there are 5 swerve teams that use swerve and have a drivebase that’s about as effective as a west coast drive.
For every one of those teams, there are 10 teams where swerve is a direct hindrance to their competitiveness and put them on Do Not Pick lists everywhere.
I’ll let you figure out on your own which tier you’d fall into.
Pros: Move and push in any direction.
Cons: If you start now, you might actually have a competition-ready drivetrain and driver in the 2022 offseason/2023 regular season. Did someone forget to tell you that swerve is HARD?
Oh, and weight, mechanical complexity, code complexity, and drivers have to wrap their minds around it.
Pro. Very hard to get pinned down by defence
Cons. Cant be heavy expensive and harder to learn how to properly drive.
But im old school driver from 07 08 and you cant beat a well driven tank drive in my eyes.
- Strafing and other swervy movements can make it easy to line up for scoring (relative weight of this advantage can be game dependent)
- Easy to drive
- Any side of the robot can be the front at any given time (opens up design opportunities)
- Provides better pushing power than other holonomic drive choices (Mechanum, H-drive, etc)
- Generally modular (can replace a whole drive unit if you have a failure)
- Complex drive (multiple points of potential failures, reliability can be an issue)
- Resource heavy (8 motors needed)
- Space allocation (modules generally must be in the corners of the robot limiting other design options)
- Difficult to use on challenging terrain (but not impossible - remember swank?)
- Cost (can be expensive depending on the design you chose)
- Weight (newer designs are getting better, but swerve can be heavy)
- Code is more complex (but not really that complex if you read the resources out there that explain the math)
I agree 100% with the statements that you should explore swerve in the off-season and get comfortable with it before using it in season (we helped one team do it during the season last year and it worked out well, but generally we discourage this). There are many good designs out there to use as a starting point or even the end point for your design and some excellent COTS swerves, so I would explore using one of them before designing your own.
All depends on your resources and what you’re trying to get out of it.
Good programming team?
Want to just learn and have no desire to put it on the field in 2020?
Think having swerve will make your robot better and win?
If you answered NO to any of the 1st three or YES to the 4th then swerve isn’t something you should be doing.
I resemble that remark!
Redacted due to poor reading
Not sure where this number comes from, and while it sure would be nice to have 8 mentors, I can personally attest that this is not the case, even for a swerve.
103 ran swerve for the first time in 2017, having never practiced in the off-season with the drive system at all. We had five mentors, only two of which had any programming experience, and did pretty well for ourselves.
Granted, we used a COTS module, but were able to successfully field a swerve bot.
Swerve is more difficult than a tank, mechanically, electrically, and programatically, and it does have drawbacks (like any drivetrain). Don’t let that stop you from learning or considering it as an option, but be ready to put in the extra time and effort to make it work if you decide to go the swerve route, and know there are plenty of resources and people willing to help here and likely wherever you are as well.
I was one of the naysayers on 103 who suggested we not switch to a swerve drive without any experience with it though I did believe it was a better solution to the 2017 game than tank. In hindsight I probably would still have been that naysayer, but, honestly, that drivetrain is a HUGE reason (if not THE reason) we were so successful in 2017 (considering our bot wasn’t much more than a drivetrain at all).
All of that being said, I still highly suggest teams have some experience with swerve before jumping into it during the build season blind.
Edit: I just had my own, “Sir, this is an Arby’s,” moment, didn’t I?
Motors, not mentors
Honestly, if teams were more conservative with their designs when implementing swerve I think the general success of their season would be more common. I saw several teams use swerve in NC this year that I think were completely unprepared for the added complexity and I just feel like they would have been far more successful with a kitbot or at least MCC.
2910 took a simple approach and didnt stick an elevator on their bot where most teams probably thought they had to reach all levels to do well. The best of the best will do everything and do it well. I think some teams think it will happen over night and that swerve might be this magical thing that makes it happen.
True, swerve is not a magic recipe for success. But there was at least one team in NC (2640) that tried swerve for the first time this year and had their best season in a long time (first blue banner since 2012 and an alliance captain at DCMP).
I can tell you that when they got their robot driving for the first time, they were pretty excited. That was pretty magical. Swervy in pink!
I expect you might see a couple of other swerve bots next year in NC given what I saw at THOR a couple weeks ago.
I would just say (and I am not the most experienced person) that it might be best not to put so much effort (and as others on this thread have attested it is a lot of effort) into just a single part of your robot. Swerve is a massive investment of effort and, as others have said, are not neccessarily going to make you a world class team. Take 254, for example. They used a tank drive and implemented a much simpler (compared to swerve) turret arm and got farther than 2910 at Houston Champs. 254’s focus on making sure they could do everything rather than just having a good drivebase was arguably a better strategy than the rather short-term approach 2910 took that won them districts. If you find a simple swerve solution, it may be worth pursuing, but a robot with good all-round capabilities is most likely better than one that has only a single but phenomenal capability.
This all heavily depends on your team’s goals (long term and short term)
I just want to make it clear that 2910 did not choose build a only Level 1 robot because we choose to use a swerve drive. This is simply not the case.
The swerve module design, fabrication process, and most of the code base we used this year was all completed before kickoff. The extra effort and focus the swerve drive required during build season was very minimal.
If swerve wasn’t a big resource drain for us during build season then why did we only build a level one robot with a single degree of freedom arm, basic hatch mechanism on the back, and pneumatic cylinders for climbing? It’s because that’s all we honestly though we could get done within our build season schedule.
I tend to disagree with you on this. Your average team will often times not realistically be able to build a robot that does everything to a high level. In most cases it is much more attainable to build a robot that can do one or two things very well.
2910 choose to prioritize a level 3 climber over scoring on the high goals. Why was it one or the other for us? We don’t have the design and fabrication ability of the elite level teams and we wanted to keep our robot simple and obtainable for us.
When choosing what direction to take it is all about analyzing the game, evaluating the different options and choosing a path that your team can actually design, build, and get driver practice with all within the very short timeline.
I am not saying you did. I am just making a comparison. One approach favors trying to go for everything, the other favors going for some of it. I have great respect for your team but am cautioning others to take that road as swerve may suck a lot of time away in a build season.
A better comparison might have been 1323 vs 254… but I digress…
Sorry, that was snarky.
For teams that have developed a stable swerve design (ourselves, 2910 and several others), there is really not much effort spent each build season to fabricate them. In fact, we start cutting metal on our modules the day after kickoff and they are usually one of the first things completed. Assembly is routine enough that we usually use this as a training exercise for newer members to learn how to assembly VP gearboxes, press bearings into bearing pockets, etc. It really is a basic component for us (just like many of the WCD teams talk about being able to “bang out” a WCD base in their sleep).
Being able to create the mechanisms to “do everything” is where we spend all of our time. Some years we do better than other years. But our ability to come up with a design that is effective at doing everything (or not so effective at doing everything) is not the result of having swerve as a distraction. It really is a result of what the students are able to dream up.
For a new team new to swerve, trying to do swerve for the first time after kickoff could present problems depending on how they go about doing it. If they pick a stable, well proven design (like 2910’s COTS Swerve, or by using the CAD from other well developed designs), they may not run into very many problems at all and it may not present much of a distraction. But we, like most other swerve teams, encourage people asking about getting into swerve to work on it in the off season and get comfortable with it before attempting it for competition. But I would say the same thing about chain in tube WCD and other more advanced drivetrain designs. It is just good practice to try to get a solid drivebase design before kickoff.
I also think this would be a better comparison. Both of these teams are leagues above 2910 in terms of design and fabrication capability etc…
Full disclosure, I do sell swerve modules and am part of a team that has had a lot of success since switching to swerve, so what I am say below could be biased.
Over the past couple of years swerve has become quite a bit more accessible and continues to make more sense for an increasing number of teams. This is an accumulation of multiple factors.
- Brushless motors are hitting their stride in FRC which has really helped mitigate the weight penalty of swerve.
- Over the past couple of years solid COTS swerve modules have become more available. Now days a team could have a complete swerve drive base assembled in just a day or two. Basically as fast as building a WCD with COTS gearboxes.
- More example code is and will continue to become available.
2910 was the only team in the PNW district who ran swerve during the 2018 season. In 2019 there were at least 8 teams with swerve (I might be forgetting one or two). If the 2020 game is good for swerve (which I give maybe a 85% chance?) I expect to see the number increase again.
All this being said, until swerve becomes basically plug and play I would still not recommend a team try swerve drive for the first time during build season. 2910 would have not used swerve for the first time during a build season. Like others have said it’s best to get it figured out in the off-season so you can reap the benefits of it during the season with minimal in-season effort.
I can’t see how 254’s turret arm is simpler than swerve… I mean have you seen their tech binder? That thing is insane.
That being said, I’m gonna have to disagree with you on this one, focusing on one or two key tasks and doing them phenomenaly is precisely what you want to go for if you’re not sure you can do it all. Good examples being 1986 in 2017, a very slow gear cycle but the best arguably shooter in the world, 2910 in 2019, arguably top 5 in terms of score generated per match, or us (1690) in 2016, we decided to give up on a scaling mechanism, allowing us to focus our attention on quick cycles through the low bar. You can also look at 5654’s robots every year, they’re always a good example of a team that knows their own limits and builds robots accordingly, always giving them plenty of time to practice and perfect the specific tasks they chose to focus on.
TL:DR: not every team is like 254 and can make a robot that does it all perfectly.
To return to the matter at hand, I believe working on a swerve drive during the off season is a good idea to help that goal, as once you’ve learned your way around it, it’s pretty much a free advantage, but I wouldn’t count on it to win competitions, whatever is on the drivetrain makes a lot more of a difference.