Following the very interesting thread on those teams that use Swerve Drives on their robots and how they perform, I am left with these questions:
Iirc, there are two Minnesota teams who tried swerve, had a ton of problems with it and decided not to do it again: 2220 in 2011 and 2169 in 2014.
We did it in 2016 (four wheel based on 221’s design) and in 2017 (three wheel custom with globoid rotation gearboxes). The 2016 ones were plagued with both mechanical and software issues but the 2017 ones ran great for the most part. This year we decided to change it up and go tank, and I am personally very happy with that decision. It has made our auto much more consistent, the driver had an easier time getting used to controlling it, and movements feel much more precise because there’s one less degree of freedom for its motion than a swerve. I don’t think we’re 100% against going back to swerve, but if we do go back it will probably be with more time invested into its development beforehand.
As you can imagine, swerve drives are considerably more complex than the basic six wheel drop center used by many teams. The complexity is expressed in mechanical (more moving parts), electrical (more motors/controllers) and especially programming (orienting and varying speed on each wheel independently). It also requires more driver practice to get the most benefits from the added complexity.
Unless you want to spend your entire build and competition season tuning and learning to use your drive base, I would highly recommend that you build your first Swerve Drive as an off season project. I have been on a team that tried to do Swerve for the first time after kickoff and we spent more time trying to get mobile than trying to manipulate the game pieces. We went back to the classic 6 wheel drop center the next year in order to have more time to work on the scoring system(s).
Also, Swerve is not particularly cheap. If it does not offer a substantial benefit to your particular strategy, it may not be worth doing every year. I’m sure there will be some that would argue that Swerve is always worth the cost, but if your team has resource constraints (budget, programmers, builders) it may not be worth it (or affordable) to you.
It is always a good idea to understand your own teams resource limitations, whether it be money, tooling or people power and to work within those constraints. A practiced driver with a good robot can often outperform the ‘perfect’ robot with an inexperienced driver.
We used a swerve drive based on an older 2767 (Stryke Force) Design last year. We had good success with it but any swerve is expensive and complex. Programming intense as well. It is also not the a great power drive for defense, with the idea being a good driver can avoid a push contest but it does take a real good driver to master. Having said all that for 2018 we didn’t choose to use swerve but I don’t think anyone has ruled it out for other years and the games that may come.
We did an off-season swerve a while back (I think it was during the first RoboRio Beta test, after we finished the Alpha test). Complicated, difficult, and ultimately the team decided it wasn’t worth it for a competition robot. I’m sure we’ll revisit it at some point in the future with a new group of kids, though!
I’m currently working on a small swerve drive for the office in order to terrorize (and bring snacks to) my co-workers.
Swerve is hard, even with ultra-light-duty hobby stuff. Hopefully this Tough Ink 115 filament can handle the lateral stresses without much flexing, otherwise I’m dropping the pursuit for a kiwi drive.
Does this count?
Unless you have a right resources, it seems like a multi year project before it is ready for a competition grade robot.
67 had a 3-wheel swerve drive flopbot in 2005. They won the World Championship with that robot, but they’ve never done swerve since.
From what I can remember from my earlier years on 67, the chatter about this robot seemed to be about how difficult it was to get right. Swerve drives were in their early stages in those times, and though it’s hard to imagine a world championship drivetrain leaving a bad taste in a team’s mouth, I think that’s exactly what happened. They’ve moved on to the familiar sheet metal skid-steer chassis since then, as it seems a lot easier for them to adapt to the game and produce quickly. Of course, I’d love to hear about this from someone who was actually on the team in 2005.
Wildstang 111 ran swerve for many many years (most seasons from 2000 to 2010, IIRC). They ran 2 speed WCD in 2011 and ran some version of tank drive since then. I believe the original decision to change away was the desire for a shifting drivetrain and not wanting to shift their swerve.
Robonauts 118 ran swerve drive from mid-2000’s through 2008, half-swerve in 2009 and then tank drive from 2010-present.
Hammond 71 ran swerve many years from mid-2000’s through 2012. If my memory serves correctly, they lost the sponsorship of Beatty Engineering in 2013 which lead to simplifying their designs significantly.
1717 used to make the best swerve drives, but they haven’t made one since 2015.
Not cool, man. 
RIP
Pwnage moved on from swerve a couple of years ago, I think.
We made a pretty cool one last season, and chose not to do so this season.
The game doesn’t always call for it (if it ever does?), so the additional complexity would have stressed our already limited student capital. We have a lot of documentation on what we learned, and how we’d do it differently, should we ever choose to do it again.
I can speak to 2016 and 2017.
2016 was 2016, lots of drive obstacles, swerve was not really ideal. Possible to get it to deal with a lot of the defenses, but not worth the effort.
2017 required a lot of different mechanisms for the gears and cubes, and between PDP slots, packaging, and focusing on the complexities of the other subsystems, swerve just did not seem worth the investment.
I’m not with the team this year, but from talking to them due to various resource allocations, etc. swerve was not something they were interested in commiting resources to. Kevin could obviously speak more to this year.
696 did one in 2015. It worked, but it had it’s faults. After seeing 1323 this year, now the students want to get back into it. We’ll see. It’s a lot of work, time, and money and there’s a lot to go wrong.
We examine the game and decide then which drive to use. The recent games have not required ‘crab’ drive. (That is the actual term used in things like camera pedestals.) We have been using our own modified WCD for several years as it is easy to design, gives the ability to shift final gear ratios, takes little room in the chassis and is lightweight. If we decided the game could required crab we would use it. We have also tried using an “H” drive once. I didn’t like it and I don’t believe it gave us any advantage.
From that, we know that crab is difficult to accomplish. Drive software is critical to it’s function, tradeoffs for weight and control system placement are needed and practice is essential. To use crab to it’s fullest, at least twice as much practice is needed. The driver must be able to see with his(her) eyes closed and visualize where the robot needs to go in a straight line. I can tell you that not all students will be able to drive effectively with crab but practice will help.
On a previous team, we did a swerve drive about 2012. Huge challenges. Never again.