1678 Swerve

[pandomains]

A few years ago, 1678 released a video detailing a bunch of tips and tricks for efficient robot design. A lot of information in that video was incredibly useful, and I hope to incorporate it into my own team, but some of it has aged in a fairly funny way.

So @Michael_Corsetto
https://drive.google.com/file/d/1DY5DO-Lea9_MrNVaOMIoRurzlg_Ew_th/view?usp=sharing

[troy_dietz]

“You should build tank style drivetrain, okay”

really doesn’t hit the same as

“Anything that moves sideways is a waste of time.”

Regardless,

[Andrew_L]

Changing your opinions based on new information over time is a normal thing.

[Michael_Corsetto]

I feel called out

This is actually a really interesting statement to reflect on. I think I’ve written about some of this elsewhere, but let me summarize the things that pushed us to swerve in 2022 for our first official season, after years of determining it wasn’t worth it.

1. Practice. We built 1 swerve for offseason 2019, 1 for at-home 2021 challenges, 3 for 2021 offseason, and then alpha, beta, epsilon and comp in 2022. That means comp, our first officially competed with swerve drive, was actually the ninth robot we built with swerve.
2. COTS Modules. The SDS and WCP modules are amazing and have improved dramatically over the past three years. We don’t have the manufacturing capacity to make our own modules like some of the OG swerve teams do.
   2a. Flipped motor is the way. Love getting that CG even lower and protecting the motors.
   2b. Wheel-in-corner is awesome for keeping our wheel base as wide as possible.
3. Falcon power. Before brushless, choosing swerve meant taking a power hit on the DT. The standard for a few years was 6 miniCIM in a WCD, teams like 1323 stepped down to 4 miniCIMs to make swerve happen. Falcons have meant we have all the power overhead we could want to the carpet.
4. Falcon integrated encoder. Having an integrated motor encoder with high accuracy means there is one less auxiliary sensor we need per module. Older motor/module combos needed 2 motors and 2 encoders per module.
5. PDH. 20 slots for motors makes swerve so easy to justify. The extra slots make designing robots a lot easier.
6. CANivore. We were right at the bleeding edge of CAN utilization on our 2021 offseason robots. The CANivore completely solved this issue for us, now we don’t worry about CAN utilization.
7. BOM Cost Limit Removal. The removal of the overall cost limit made a lot of the above possible.
8. Code Examples. We went through four different swerve code frameworks before finding one that we were able to leverage and get the teleop control and auto path following that we wanted. We’re still not perfect on this front but it’s getting better.
9. Flat field. Pretty self explanatory.

All of these factors, in my view, pushed us to swerve in 2022. We’ll see what 2023 has in store!

Best,
-Mike

[s-neff]

I wouldn’t even change the workshop script either. Build tank, buy swerve.

[Richard_McCann]

At 2019 Madtown as the MC/GA I called the 1678 bot the “Heresy”…

[Richard_Wallace]

@Michael_Corsetto

Flat field. Pretty self explanatory.

On this field, anything that DIDN’T move sideways was left behind.

Game designers noticed that.

[saikiranra]

@Richard_Wallace

Game designers noticed that.

If the implication here is that we won’t get a “good” swerve game for a while, how do you know?

[Richard_Wallace]

[Tom_Bottiglieri]

This is why I have been preaching that teams should use swerve since at least 2014. I’m glad I don’t look as silly as Mr. Mike does here.

[R.C]

Yeah Tom we followed your advice in 2015 and looked silly Insert @Eugene_Fang gif

[josesantos]

@s-neff

Build tank, buy swerve.

I’m dreaming of a world where the options are buy swerve OR…

Just stick to the (totally-hypothetical) AndyMark x REV brushless-powered WCD Kitbot based on NEOs and ION extrusion.

@Andy_Baker @Greg_Needel please

[Eugene_Fang]

ImHelpingISwear.gif

[Richard_McCann]

Is that from 2015 or 2003?

[Richard_Wallace]

@Richard_McCann

2015 or 2003?

First time, the GDC, tried to make us stack totes by strongly discouraging defense. Second time, they tried to make defense impossible.

Both attempts failed. Neither game is recalled fondly.

[Anteprefix]

For those who would like some more pixels (and audio too):

Q0068

[anon37142155]

@Michael_Corsetto

2a. Flipped motor is the way. Love getting that CG even lower and protecting the motors.
2b. Wheel-in-corner is awesome for keeping our wheel base as wide as possible.

How big an impact did you find switching to the flipped (MK4i) from the non-flipped (MK4) have on your drive?

Edit: in the past they’ve said that they’ve tried a bunch of different COTS swerves, but ended up liking the MK4is the best.

[Michael_Corsetto]

@anon37142155

How big an impact did you find switching to the flipped (MK4i) from the non-flipped (MK4) have on your drive?

“How big an impact did it have on our drive?” is a hard question to answer, we like them more for a variety of reasons. If there are specific metrics/aspects of the differences your curious about let me know.

[anon37142155]

I’m more curious about a few things

1. Failures/damage: I’m not sure how much you used the MK4s, but did you find the MK4is to have less failure/ parts breaking. Also if other teams have damaged either of the modules.

2. Stability: I know there’s software tuning you can do to improve stability, but would got say there was a significant shift in stability from the MK4s to MK4is

[wgorgen]

@anon37142155

I’m more curious about a few things

1. Failures/damage: I’m not sure how much you used the MK4s, but did you find the MK4is to have less failure/ parts breaking. Also if other teams have damaged either of the modules.

I’ll chime in here. Our swerve modules have a similar, upright form factor to the SDS MK4 (and earlier generation) modules. In 2019, we had several matches where we were pushed into the cargo ship by a defender. Since the cargo ship had the cutouts to allow bumpers to pass under them, our swerve drive motors were the part of the robot that ended up making contact with the cargo ship. We damaged 3 motors at CMP in 3 different matches (NEO motors - the cover failed and fell off rendering the motor unsuitable for continued use) and bent 2 motor plates from these impacts. We have not suffered similar damage in either 2020 (1 competition) or 2022 (6 competitions). However, we did have one of our motors get unplugged in 2020 that may have been the result of another robot entangling with the exposed wires.

Several teams that use upright modules have incorporated shields to protect their motors. We did something similar in 2019 after CMP for our offseason events that worked pretty well, but we did not carry that forward into either our 2020 or 2022 robots. Here is a bicture of the corner shields that Team 1690 added to their robot to protect their upright modules:

From a non-damage/failure perspective, we went with a full width over the bumper intake for 2022. Having the swerve motors sticking up at the corners of the robot blocked some of that width. It turned out to be a relative minor problem due to the overall size of the game element this year. But it would have been nice to have a completely flat surface on that side of the robot to be able to intake the balls across the entire width without having those balls hit the motor on their way into the indexer.