Team 254 Presents: 2022 Recap, Technical Binder, and Strategy Presentation

[Torrance]

Team 254 is proud to present our Technical Binder and Season Recap Video for the 2022 season.

Match strategy is an integral part of FRC competitions, but many underestimate its importance. This presentation summarizes our approach to match strategy as we entered playoffs at worlds, and helped guide our scouting meeting with Team 1619 prior to alliance selection. We hope you find it interesting!

I am also proud of the students’ presentation for FIRST Updates Now’s Behind The Bumpers video.

At this time we are not ready to release our Code. Also, our build blog took a different form this year, more of a short-form series of a bullets rather than the nicely formatted blogs. This was easier to produce and more useful to our own students/internally, but is less useful to share publicly.

If you have any questions about the robot, Technical Binder, Recap Video, or Strategy Presentation, please feel free to ask!

[Ian_McTavish]

Thanks for this. Really appreciate you sharing this. Very cool to see the strategy presentation.

[Alex_Y]

I wouldn’t want to score cargo in the high school

[eckshn]

Is the reason why for path generation Quintic Hermite splines are used because of the fact it’s determined by points and a derivative (velocity) at each point? Or was there a different reason as to why that method was used compared to other spline generating methods?

[anon37142155]

From your technical binder, it says that you decided on a fast agile swerve, was this decided immediately after seeing the flat field or was this chosen later on?

[Jonathanshani]

I must say that every year I get excited reading your tech binder and this year is no different, but looking at this bot leaves me in awe. The design looks impecable, like every bolt and standoff are in exactly the right place. It seems that as I get more “Real world experience” in mechanical design, the way you design your bot impresses me further.

With this I have 3 questions about things that I want two make sure I got right about your design.

1. Is the shooter assembly only connected to the turret with 4 bolts? If so what size bolt did you use, and what percausions did you take in the design process to make sure that it is stiff enough and durable enough to withstand the vivrasions and kickback of the shooter?
2. I would love to understand more about your serializer design, mainly the integration to your chasis.
3. If there are any pictures of your battery inside the robot, It would be great to see, as I don’t fully understand how it mounts there.

Thanks a lot for sharing this document every year, it’s always such a good learning experience for our students and mentors!

[MaxShankland]

In the binder it says you modified the swerve modules to add bearings that support motor shafts. Is it possible for you to post pictures of what this modification looks like along with information on the bearings you used?

[UnofficialForth]

@Torrance

At this time we are not ready to release our Code.

Do you plan to release it/have an estimate of when that will be?

[Andrew_L]

I liked seeing the solid bellypan. It’s something I’ve considered for a while now. Were you happy with the 1/8" solid aluminum pan? Going back, do you feel the weight it took up limited your design approaches later on? Looking back, would you have changed the thickness, or lightly pocketed it for weight purposes?

[jacob6838]

What is holding the elevator carriage on the sides? Are there some recessed bearings or something?

[Torrance]

This was largely decided immediately due to the combination of the flat field and the removal of many of the previous barriers that Mike Corsetto excellently outlined in this post.

The swerve was absolutely the right decision for this game, and now that we feel confident in it I foresee us using it in as many future games as we can. Moving Sideways is no longer a waste of time.

[josesantos]

Is this a genuine Kaydon (Reali-Slim?) bearing? Regardless, would you mind sharing the source/vendor?

[AarnavS]

Hi, This is Aarnav, software lead on Team 254. The main reason we use Quintic Hermite Splines is because the curves are twice differentiable and thus keep acceleration continuous (compared to jerky, non-continuous acceleration of other spline generation methods like cubic hermite splines). They also minimize curvature between splines (which creates smoother curves and less wheel slip). With a computed velocity, acceleration, and position at each point of the trajectory, driving Quintic Hermite Splines allows us to utilize feedforward control based on the precomputed velocities in addition to positional feedback control.

[Torrance]

Shooter Mounting
There a 4 CNC milled adapter blocks that connect the stationary main sideplates of the shooter to topside of the rotating inner ring of the turret. The blocks have 2 #10-32 cap screws that go into the turret ring, and 1 #10-32 that goes through the shooter plate and taps into the block. We didn’t do analysis except back-of-the-napkin “feels about right” math. The screws are well spaced out and the clamping force and friction generated by 10-32 screw is quite high. More screws probably would’ve helped a bit but after we tried this method in 2020 and it worked fine we felt comfortable doing it again in 2022.

Serializer

Functionality and Ball Path:
The Serializer and Feeder subsystem funnels balls from full-width intakes to a single ball stream up into the Shooter. Wheels made of stacked 6" diameter polycarb plates with grip tape on the outside funnel the balls. The ball is then pulled in and up into the shooter by powered top rollers, with a combination of grip tape or heat shrink to have the right amount of grip.

Powertrain:
The top rollers and funneling wheels for the robot’s left side are powered by a single Falcon 500 with a gearbox, bevel gears, HTD belt runs, and a connecting jackshaft. The top rollers are 1" OD 1/16" wall tubes with 3D printed endcaps with pulleys and bearings in the ends spinning around 1/4" OD deadaxle standoffs/shafts. The outermost top rollers were passive and just held the ball from popping out the top when being funneled.
A mirror image gearbox powers the robot’s right side wheels and top rollers. These two motors let us drive just one side or other, allowing us to feed one side into the shooter first, or shuffle a ball from one side to the other.
Shuffling was needed to open space for the incoming 2nd ball, which may be of the wrong color so we needed to be able to pass by and go out the shooter while not touching the correct ball we are already holding.

Structure:
The heavily pocketed horizontal 2x1s hold the funneling wheels, gearboxes, and bent-polycarb shelves that the ball sits on as it gets funneled.
The large triangular polycarb sideplates hold the top rollers, Color Sensors (REV V3 worked well) and Banner beam-break sensors. It also held the standoff that large curved ramps could pivot on, to be lifted up for robot maintenance or getting the battery. These polycarb plates also held a piece of surgical tubing that served as a reaction force to push the ball into the feeder top rollers. The surgical tubing was lightweight and worked in both directions, always complying to push the ball against the roller it needed to. This clever bit of design is one of my favorite features on the robot.

Mounting:
3D printed (Markforged) blocks connect from the top of drivetrain framerail 2x1s to the horizontal pocketed 2x1s of the Serializer. They have a little “jog” to them to allow mounting to frame rails outside of the MK4i swerve modules, then plug into the endcaps of the horizontal 2x1’s at the correct spacing the 6" funneling wheels wanted.

Square-shaped 10-32 nuts are pressed into the 3D printed blocks. I do not recommend this, this offseason square nuts spun free in the pocket and we had a real bad time getting the Serializer off. We had much better success in other locations form-tapping 3D prints and just threading directly into the plastic.

Battery
The battery is in the direct middle of the robot. This had some weight/CG benefits, but mostly was dictated by packaging. During initial placement we had a large color sensor under the bent polycarb cross-bridge pieces, this prevented the battery from being near the frame rail on left or right side. The elevator and stinger for climbing prevented the battery from being near the frame rail on the front or back. The middle was really the only open spot.

The battery is held in with a 1/4" thick aluminum plate held on standoffs, with 2 webbing straps with buckles pulling it down to the bellypan, like we’ve done for battery boxes in the past.
A 3D printed ramp on on side holds the Anderson connection and the wedge ramp allows the battery to angle up and slide out of the robot. The large curved ramp of the Feeder needs to be lifted and pivot up to get the battery out.

[Torrance]

We green-loctite glued 19mm OD, 10mm ID flanged bearings (4390N154) into the holes (19.13mm ID) in the top swerve plate. A 3D printed adapted slide over the end of the falcon shaft and into the inner race of the bearing. A flanged button head cap screw on top threaded into the end of the Falcon shaft and held everything together.

We don’t really know if this was necessary or helpful. Plenty of teams ran MK4i all season and offseason without this addition and had no issues. This didn’t seem to introduce any new issues for us and was a pretty small cost on weight, so we kept it. Not sure we feel we need to do this next year.

[Torrance]

The 1/8" bellypan may be the most contentious decision we made early-on in the season. The low CG and having the option to add extra mounting holes for electronics or a redesign of the elevator gearbox were fantastic. Minimizing the amount of ball fuzz and carpet dust we pulled up into the robot made it stay cleaner. Although not being able to spray air and push dirt out of the bellypan made it hard to clean.

I don’t think we would change the thickness, having the stiffness with the battery in the middle was just barely there at 1/8" thick. However, I honestly think we should have pocketed it because we were fighting weight the entire season. Robot weighed 124.9 lb at Worlds and we were pulling every trick in the book to shave weight. Pocketing tubes, carbon fiber rods, low-infill 3D prints. It was exhausting and a headache the entire season.

[Torrance]

The carriage uses the new WCP elevator bearing blocks (WCP-0323). These blocks have very small (12mm OD) bearings that roll against the inner faces of the elevator uprights and constrain the carriage side-to-side.

Luckily the side-to-side loads weren’t that high this year, mostly front-back like most elevators, we did eventually replace these tiny bearings in the offseason but they held up through Worlds.

[Torrance]

We purchased the bearing from Ebay, which allowed us to get it for much cheaper than buying it new from Kaydon directly. I think we actually ended up with a FAG-branded bearing, but it doesn’t really matter since almost any large thin-section bearing (including much cheaper knockoffs) would be fine for this application with such low load and stiffness requirements.

[Ethan_Reed]

Just to share another way to get these as @josesantos brought up a good question here. 118 struggled a little to get our hands on large thin section bearings last season. We used Lily Bearing for our turret bearings. We bought ours for just under the $600 limit and got them just in time to stick them into our robot to film our reveal video, after about 6 weeks of shipping from China.

We had no issues with the quality of these Kaydon clones and will likely use them again. We will continue to plan for long lead times and order early.

Also, huge thanks to 254 for putting together these resources. I always look forward to the technical binder and the CAD screenshots/descriptions in these threads.

[waialua359]

Based on personal observations, I feel like Swerve helped you and other folks a lot in one area not on the list, vs. it being a “waste of time” and using WCD again.

Avoiding (limiting) defense.

SDS swerve defensive robots with good drivers vs. WCD driven robots were deadly.