Opinions on Swerve Drive

After viewing the other threads about who is and was using swerve drive, I couldn’t help but wonder what everyone’s opinion on swerve is. So what do you think? What are the advantages or disadvantages?

I imagine the responses will be fairly similar. My take on this issue is that swerve is a highly complex mechanism, both mechanically and programatically. If a team has the right resources, and can dedicate an entire offseason to perfect the swerve hardware and software, swerve drive simply ends up being another tool to use in case the game truly requires omnidirectional motion. However, swerve is hard! There’s a lot of things that could go wrong, and designing a great module is very difficult.

Furthermore, since the COTS swerve options are limited, it means that you have to manufacture swerve in-season, which could potentially eat up a large bulk of your time.

Swerve definitely has its advantages if a team can dedicate its resources to getting the best possible modules and software they can. Often, teams can’t, and moving sideways isn’t really necessary in a lot of games, so 6 wheel WCD often ends up being the best option for teams.

In my opinion, almost 100% of robots that have swerve would be better without it. It’s added complexity without a big enough return on investment. Even the best teams who execute swerve often struggle with breakdowns at the worst times. I’m specifically not mentioning any teams, because I don’t want to hurt any feelings or take shots at teams that are really quite excellent, but there are several great teams that could dominate if they stopped building swerve bots.

My hot take is that swerve drive* offers almost no competitive advantage that an equivalent amount of good driver practice can’t also provide.

*Or any other mechanism of your choice.

With no driver practice, swerves offer a good bit of an advantage over WCD.

With a reasonable amount of driver practice, swerves don’t really offer a significant advantage over WCD.

With driver practice and experience over multiple years plus incredible code (see 1323, 16, 1640, 1717) swerve offers a much more significant advantage over WCD.

Driver’s skill ceiling is higher with Swerve IMO. However its much harder to get it to that point.

Teams that do swerve all the time show how good the drive train can be.

Most teams don’t get anywhere close to their skill ceiling even with a 6wd.

As someone who has personally designed, built, and drove both swerves and non-swerves at the FTC and FRC level, I would say that the swerve robots left a more lasting impact on the students on the team, was more inspirational to the students and parents that we interacted with, and served us better on the field because of our sound situational understanding of swerve’s strong suits and our willingness to dedicate off-season time developing it.

However …

Very seldom do I find that the drive train is a deciding factor in how well I think a robot can perform. With just about every drive train, I’ve seen teams that know how to use it optimally and hence perform very well and consistently with that. I know teams who have the best looking, on paper best performing drive trains but simply misuse it and struggle. Simply put, If a robot is good at the game, it is good at the game.

This year for example, while speed and efficiency are critical, these areas are much more important in the context of delivering the power cubes. How quick, consistent, and smooth is your elevator/shooter? How high can you reach? How many cubes do you drop? How much control do you have over where the cubes get placed? These are much more critical from a scouting standpoint than “how many feet per second can you drive and in what directions?”

It’s hard for me to consider one drive train outright better than another. I would say that more teams would be better off staying simple and working within their team’s resources. That being said, I’ve had much experience working on and conversing with swerve teams whose sentiment was that it would be harder and more costly for them to switch to another drive train for any given year (as counter-intuitive as that may sound).

It is important for teams to seriously consider what drive train they want to use, because as I mentioned being able to utilize that system’s full (or close to full) potential is critical to a team finding success. However, blanket statements about the effectiveness of one system over another is over simplistic and not very helpful in my opinion.

Mecanum drives however, instantly get put on my do not pick lists :wink:

Swerve is only good if you know how to use it. I’m always disappointed to see a swerve robot drove like a tank. Swerve when properly used can offer a huge advantage over tank, especially in defensive maneuvering. But to become good at swerve, just like everything else, it takes a ton of time.
My team is personally not going to make a swerve drive anytime soon. The way we see it, we would rather work on making lighter and stronger WCD style robots that are more versatile. 254 has won 3 world championships using a WCD, and countless blue banners.

I suggest you take a look at 230 this year. They most certainly should not be on anyone’s do not pick list (and they’re the ones picking usually anyways).

  • It’s a great learning experience, both in the mechanical and controls disciplines.
  • It’s really cool when it works.
  • It’s not very cool when it doesn’t work.
  • It may offer some advantages during matches.
  • It’s an engineering tradeoff like any other.

It’s my favorite drivetrain but we haven’t yet used it in-season. Just know what you’re getting into. I do partially subscribe to the mentality of ‘WCD is tried and true’, and we did a WCD this year. But I don’t think it’s fair to dismiss any drivetrain so quickly. If you have the resources, I would highly recommend it as an offseason project. We learned a lot from it, and plan to learn even more this offseason. Based on your own experience you can make the right decision.

My ‘custom user title’ is appropriate…

A good swerve drive is a beautiful thing to watch. It’s the reason Bomb Squad (16) is typically on my list of favorite robots each year. As a Northwesterner I really enjoyed watching Jack in the Bot (2910) this year.

The same goes for 1986 in 2017. Like I said, the importance is in how well you utilize and understand your drive train choice and less about what your choice actually is. I don’t actually put mecanum drives on a do not pick for the sake of a vendetta against mecanums (sarcasm on the internet is tricky), and I hope my post was clear in conveying why I don’t do that.

As a side note: 230’s pac-man ghosts on their robot that change color if they have a cube is one of the coolest things I’ve seen this year. Whoever came up with that and actually followed through with it deserves kudos.

This. I programmed one swerve drive as a senior, and it was by far the coolest thing I’ve ever successfully done in FRC as a student. The drive drained resources that might have been used to improve our game piece manipulator, and thus likely hurt our competitiveness, but I believe that my teammates and I benefited from the experience of building a swerve drive.

TL;DR: Make one during off season if you can. You probably shouldn’t compete with it, but you will have made one of the coolest things you can in FRC.

Lots of negative vibes here. 1640 since 2010 has done swerve except 1 year and that was a Breacoflex belt tank. Swerve has forced our team to become much more organized and focused on project management (time management). Yes, swerve is a load. A serious load. But,I as a MENTOR can not think of a better year to year mechatronic study. Swerve allows me to push a group of students beyond their perceived ability. It allows me to push the concepts of automation to our students every year. I do not place as much value on winning as I do the personal growth of our students. Swerve makes them grow. The skills acquired from swerve has also lead to us being pretty good at winning too! So if you want to join the pivot club think hard about your teams ability and if you want to push hard and make the changes to be successful. If you do not have the resources and dedication do not do it.

Thanks for not calling us out!

-3419

In all seriousness, we’ve done swerve for the past 4 years. I’ll let our track record - positive and negative - speak for itself:

-Recycle Rush: Swerve drive was effective for our strategy which involved lining up with tote chute (other bots were making endless K turns). It never failed us throughout the season. On the negative side, it was quite heavy, so we had no weight available for can-grabbers.
—NYC regional: Ranked 1 and lost in the quarterfinals due to a non-swerve failure, and we won the Innovation In Control award.
—DC Regional: Ranked 1 and won the tournament, and won the Quality Award.
—Championships, we performed pretty well but lost in the first round of the playoffs

-Stronghold: Swerve drive was critical for getting balls and lining up obstacles. We went with a 8" pnuematic wheel, but still fit under the low bar, which was a challenge with swerve, but we made it work. It failed once in the season, but in a way that a typical drive train would have also. That failure didn’t cost us anything.
—NYC Regional: Ranked 3, lost in the finals. Won the Quality Award.
—Tech Valley Regional: Ranked 3, lost in semis. Won the Creativity Award
—Championships: Ranked 46 but was 1st pick of 6th alliance. In the playoffs we faced off against 1114, 27, and 20 and shut them down with great swerve-defense: Quarters 4 Match 1 - Carver Division 2016 - The Blue Alliance . Eventually lost to the world champs in the semis.

-Steamworks: Swerve drive was excellent for getting gears and manuervering about the field. It failed terribly in the HVR semifinals, costing us that tournament.
—Hudson Vallery Regional: Ranked 3 but list in playoffs due to swerve failure. Won Innovation in Control Award
—NYC Regional: Ranked 2, won.
—Championships: Ranked 6, lost in quarterfinals

-Power Up: Our best swerve drive yet, but our robot was otherwise not very good this year. An electrical issue cost us one game, but other than that it worked quite well.
—Hudson Valley: Ranked 16, lost in quarters. Innovation in Control Award
—NYC Regional: Rank 4, lost in quarters. Quality Award.

From a purely competitive standpoint, I’d say there are games where swerve drive is hugely useful (last year), and others where is was not (this year). The benefit of being hugely manueverable need to be weighed against the drawbacks of complexity, size, and weight. We’ve done a pretty good job of minimizing those things over the years (this year’s module was 5.5" wide x 8.25" long x 10.75" tall and 8.0lbs, motors included), but it’s still a tradeoff that needs to be considered.

From other standpoints, we think the swerve does a great job of inspiring students. Our students take a lot of pride in it as our “team signature”. And it has certainly drawn a lot of recognition, with 7 engineering awards in the past 4 years.

1 Like

In Wisconsin Regional this year watching Roboteers 2481 and Lights Out 323 was amazing. Originally, I thought this game wouldn’t matter that much what drive train a team had. But the swerve does everything better, faster, smoother.

Increased mobility. That’s what those guys have that our AM14U didn’t. They could just start moving in the desired direction whenever they decided to, where we had to turn our robot that direction first. It made smooth, fluid and fast movement.

That being said, my kids still build a robot with drills and saws, welders. Mostly you can’t make swerve without NC tools. You have to make many modules that are identical. You Must build a practice robot because you’re not going to get much practice before stop-build. Just too much work for my team and our crude tools.

Though I’ve been intrigued by some of the more basic swerves that have shown up on CD. I do think it can be done. Just uneasy to ‘jump in’.

COTS swerve are interesting too, but they’re $1.2k+ for the 'bot and $1.2k again for practice robot. If I had that much to spend, I’d be buying an NC router.

Not sure if my team will ever get there. Sure would be fun.

Ditto’s to most everything said here, very thoughtful and honest, because every team has it’s own context.

However, from our perspective swerve is not a part time endeavor…a maybe next year if the game favors it, or something else if it doesn’t. We believe you are all in or not. Those teams that have committed to long term continuous improvement of swerve may likely tell you they have to invest as much energy into not using swerve, as those trying to convert to it.

Now granted we are biased in our own context, and we too recognize when swerve has it’s advantages…but we would struggle to find a season when it would be an outright disadvantage. I know many of you are just waiting to point out the power and traction thing, but in our world, if you are designing swerve for a pushing match you are kinda missing the point.

The overwhelming majority of the parts on the 323 machine could, with some effort, be made with manual tools (and a 3d printer, but those are common enough imho) The main bearing plates could be done on a lathe or, with a rotary table on a manual mill (that would be my preference, I don’t like lathes). It would take a bit of work, probably finding a sponsor to do it for you or investing in the machines would be a better use.

The real problem with swerve, imho, is cost. For a single module you need 2 or more ESCs and 2 sensors (the CTRE encoders are nice).

This year’s game is really well suited for swerve in my opinion. The field is fairly closed off with lots of alleyways and tight corridors to get through and often you are trying to pass one of your alliance partners in those areas or trying to get around a member of the opposing alliance. The extra maneuverability of swerve makes these areas of the field easier to maneuver in.

When we you get more than 5 or 6 power cubes on the scale, it gets tricky to get additional ones placed without knocking one down or having yours fall. The ability to strafe sideways slightly to find a nice level spot to place the cube certainly comes in handy.

I have watched a lot of teams with 6 wheel drive bases struggle to line up on the hang bar and need to take several runs at it to get the alignment they want. With our swerve drive, we are able to adjust our position in all 3 axes (lateral, fore/aft and rotation) with ease.

Probably the best showcase of swerve this year is moving the pyramid of cubes into the vault. Our team can put the control in field orient mode and then translate back and forth between the pile and the exchange while rotating in between. It is so fast and smooth! And when we are working with an alliance partner, we can set up next to the exchange while we wait for them to push their cube through and then translate sideways to the opening as soon as they pull away. We have found that when working with a partner with a traditional 6 wheel base, we can usually cycle 2 cubes for every one that they are able to cycle.

Swerve may not be the best choice every year, but this year seems particularly good for swerve.

Interesting choice. :wink:

I will invite you to reconsider after match 42 on Turing this coming week.