FRC 2357 - System Meltdown - 2022 Build Thread

[coderkevin]

Welcome to the System Meltdown 2022 build thread, presented by #openalliance.

This is the first time our team has participated in any kind of regular build thread. We’re going to give it a try and see how far we can get with these updates.

This year will be a special challenge for us (as I’m sure it is with most of you all too). We’ve only got about 12 students on our team compared to our normal 25 and so many of them are freshmen/first year students. But our captain is a seasoned senior and we’ve got some great mentors to help us out, so we’re going to show our first years how it’s done!

System Meltdown as a team has been around since 2008, but over the last couple of (in-person) years we’ve really been raising our engineering level and our game strategy. In 2019, we got our first win at a regional as a second pick and went to champs for the very first time. In 2020, on week 1, we were 2nd pick of the finalist alliance and received a wild card to go to champs again (which of course didn’t happen).

Anyway, all that to say, we’ve always been a very open team. We like to default everything we do to open source in the spirit of community and coopertition! Feel free to ask any questions as we go and we’ll answer to the best of our ability!

This year we’ll be competing at two regionals: Greater Kansas City, and Central Missouri. We decided to keep it local mostly due to covid and travel issues.

Good luck everyone! And hope to see you out there!

[coderkevin]

Here are some links to follow along with us! All of these links are live and will stay updated throughout our build.

Build/Competition Schedule:

2022 Build Season Schedule

Product Breakdown Spreadsheet (BOM/parts tracking):

Product Breakdown System - Rapid React

Main Robot CAD:
https://a360.co/3GfXZsy

Code repo:

GitHub - frc2357/FRC-2022: Repository for FRC Team 2357 for the 2022 year: Rapid...

[coderkevin]

Post-Kickoff Update:

What a great kickoff! We had 10 students physically present, and one via Google Meet. We also had several extra adults (previous mentors and alumni) to help us with game strategy.

First impressions on Rapid React:

1. Less complex than previous years
2. The middle of the field is going to be utter chaos
3. A little surprised it’s another shooting game
4. The TERMINAL looks mostly useless except for the least experienced teams
5. This game is going to get very polished and competitive quickly this year

For our team, we decided our goal this year is to be a solid 1st pick at any regional. We’ve always been a 2nd pick for elims, but we believe we’re ready to move up, and this year is the year to do it.

Here are the goals for our robot this year to be a solid 1st pick:

1. Solid, fast climb to Mid rung. Stretch goal: Transverse climb for the RPs.
2. Shoot high goal from anywhere up to 15 feet away.
3. Auto: Shoot 3 in high goal. Stretch goal: 5 in high goal for quintet.
4. Teleop: Score at least 8 in high goal each match.

Drive base decisions: We’ve already decided to go with our chain-in-tube WCD from 2020 with some incremental improvements on the tensioners and press-fit bearings instead of bearing blocks. 30" long by 29" wide. Driving 6 4" custom 3d-printed/aluminum plate wheels. We may use omnis on one side and/or increase the center wheel diameter for a simulated drop center for turn scrub. We’re driving it all with custom gearboxes using 3 Falcons on each side in a standard single-speed, two-stage gearbox for a speed target of around 12 feet-per-second on the floor.

Some notes on the above goals: We are definitely being ambitious this year especially with how many first-years we have on the team. However, we’ve been training them since last summer and their proficiency levels are getting there. Two of them can already run our Omio X8 router and they can do some CAM. Another one will be helping with sub-assemblies on CAD this year, and a couple others with 3D printable parts. Also, aside from the climb, all of the tasks required, we’ve done fairly well in 2020. We had a limelight-driven turret then that was fairly accurate, as well as a decent intake and solid drive base. So we’ll iterate on what we learned from then.

One last thing: By our calculations, we will need 16+ feet of ceiling clearance to practice high hub shots. Our practice carpet is set up in the cafeteria near our shop and the ceilings there are just not high enough (we were hitting ceiling tiles with our 2020 shooter regularly). So we’re clearing out our entire storage area for our shop as it’s the only place with ceiling high enough for us to practice! It will be a tight fit, but we’ll make it work.

[coderkevin]

Here are some diagrams on what we’re planning for a climb. Some details still to figure out, but we’re looking at two sets of hooks: Static hooks set about 30 inches above the frame, and then long pivoting single extension arms. Here’s how they’ll work:

Step 1: Extend up to mid bar:

Step 2: Pull up on the mid bar. When the long arms are fully retracted, the static hooks lock into place on the bar.

Step 3: Pivot the long arms about 30 degrees with small pneumatic cylinders, and extend them to be under the high bar.

Step 4: Pivot the long arms back to center with the pneumatics, and then retract to drag the arms against high bar until the hooks lock into place.

As the long arms continue to retract, they pull the robot forward and slide the static hooks off of the mid bar.

Step 5: The robot swings forward on the long arms as the long arms continue to retract.

Step 6: Pull up onto the high bar until the static hooks are set.

What we like about this climb:

1. It only has two degrees of freedom
2. We think it can be pretty fast

Some concerns:

1. We don’t trust 1/16" aluminum to take torsion loads, so we’ll use 1/8" square aluminum tubing instead.
2. These uprights are going to block shots from our turret, so we need to figure out how to get them out of the way.

[Landon_K]

@coderkevin

We don’t trust 1/16" aluminum to take torsion loads, so we’ll use 1/8" square aluminum tubing instead.

Assuming the robot won’t rotate during the climb, I don’t think the set of hooks should take any torsional loads. I think the majority will be tensile in nature with some bending moment during the transitions.

I really like the idea conceptually and think it should work!
Something to consider: I think the largest concern would be releasing the static hook from the previous bar. The ‘pull up’ angle has to be less than the hook tip angle or there will be some interesting pulling stresses going on as well. Coupling this with the fact the robot will likely be swinging, this could be a delicate balance for a hook shape.

[coderkevin]

Thanks for the comments! We’re thinking while there should be any twisting motion, there will still be forward/backward rotational forces, which is what we’re concerned about. The slightest bend in the extending pieces could cause it all to bind up. Good observation on the hook shape. I expect us to go through a few iterations on hook shape for sure. We do have a backup plan to add small pneumatic “kickers” to eject the rung from the hooks if needed.

[Nick.kremer]

I really love the way you sketched the robot reference frame with the robot extension and height limits during each stage of the climbing sequence. Definitely going to be stealing this!

[coderkevin]

Another Update!

The prototypes are coming along:

Intake: Like some other teams, we’ve found compliant wheels to be a good first contact with the cargo balls. We started with 4" wheels, but will try out some 2" wheels to see if they work well enough. For our second roller, we have some of the ThriftyBot vectoring wheels on either side with some stealth wheels in the middle. It worked pretty well against the bumper, and this Saturday we’ll mount it to a drive base to give it a full drive try. Promise I’ll get some video for that!

Shooter: We tested out some different amounts of compression on a hooded shooter. 1/4" is definitely not enough, but even 1/2" works fairly well. We used our 4" neoprene rollers that we used in 2020. This is driven with a 1:1 gear ratio on a single full CIM. We’ll try more next Saturday with 2 CIMs and some tighter compression to compare. (Please excuse the mess, we’re moving all of our storage out of this space so we can use it for drive practice, we need the higher ceilings!)

2021 Shooter Prototyping

We’ve also done some more detailed layout for the climber, and ended up working on some packaging layout for the rest of the robot in the process.

We’re planning on the turret being able to shoot from any direction at angles 65-85 degrees. We determined shots at less than 65 degrees look pretty rough so our plan is to be able to shoot from up to 15 feet away at 65 degrees and then all the way up to a fender shot/layup at an 85 degree angle. This is why we’re using a 2-stage extension, to keep the height of the climber under the turret shots.

Programming Update: We’ve got the new WPILib on the programming laptops, and our programming lead is trying out the latest trajectory code. We’re also researching some vision solutions for recognizing cargo by color and shape. We want to be able to “target” a cargo visually and have the robot drive right towards it. Right now we’re thinking one of three options for this:

1. Pixy CMU Cam. It recognizes color well, but we’re not sure if it will give us the shape information we need. Also, it can’t show us video easily.
2. Photon Vision. We haven’t tried this out, but we’re going to see how it goes.
3. WPILib PI and custom Open CV python code. We did an offseason project using this environment and got pretty comfortable with it. It might be nice to have 100% control of the Open CV code and the video, and we could even paint the targets on the video feed, which would be a real plus. We’d also have the ability to split off the video to log it, etc.

Anyway, hope the build season is going well for all of you! Cheers!

[coderkevin]

End of first week, second weekend update:

• Finished shooter prototype and testing
• Finished intake prototype and testing
• Finalized drivebase CAD
• Started superstructure CAD
• Started cutting drive base on CNC router

The intake prototype tested pretty much as we had hoped. We’ve decided on a 2-roller design: the lead roller is 2" low durometer compliance wheels, the secondary roller is 2" ThriftyBot vectoring wheels, with 2 2" stealth wheels right in the center. The effect of this is the first roller grabs the cargo effectively, and the vectoring wheels slide the cargo ball along the bumper to the center, where the stealth wheels pop the cargo ball up above the bumper.

2022 Intake Prototype

The shooter prototype yielded some interesting results. We tested a hooded shooter using 4" diameter neoprene rollers, two of them, each 1 inch thick. They also had safety wire around them to prevent delamination at high speed. We tested compressions from 0.5" to 1.25" and we found 1" to be the sweet spot for this setup. We were testing using 2 CIM motors at full PWM speed on a 1:1 ratio. The resulting height was pretty good and we were getting the right height at about a 10 foot range. Our goal is to be able to shoot anywhere from 15 feet though. In order to accomplish that, we will be using 2 Falcon 500’s to power our shooter, which will give us some more power for the shooter, and we may also gear up the shooter wheels if we need to.

Additionally, another thing we tested is we lowered the PSI of one cargo ball to 2.5 PSI, and raised the other one to 4 PSI. There was a very noticeable difference, one felt very “squishy” and the other one felt much more “firm”. However, through our testing at 1" compression, both seemed to launch and land in a very similar position, despite the large pressure difference. Very encouraging!

2022 Shooter Prototype

Last in this update, we have completed clearing out and re-organizing our storage area into our new practice area. We’re packed to the gills elsewhere now, but we’re super stoked to have our carpet down in this area. It’s a bit smaller than 1/2 field, but it gets us the height we need. For reference, the bottom of our lighting fixtures are 12 feet high and yes, we’re a bit worried about hitting them. Glad they’re LED now instead of fluorescent tubes! Anyway here’s the before and after picture of the storage area to practice field conversion:

Before:

After:

We’ll have more updates this week. Expect to see a post from one of our students on our drivebase design, explaining the construction and features. Until then, cheers! And happy build season!

[coderkevin]

Here’s a post from our CAD student, @CT-7567

Hi everybody this is team FRC 2357 System Meltdown’s update on this year’s Rapid React drivebase. We are using what we have learned in 2020 to hopefully make our most effective custom west coast drive base yet. To do this we are using features like…

• A chain in tube design with 1 simple compact tensioner per chain (UMHW rod in a slot with adjustable screw on top). The reason we are using tensioners is so we can press fit our bearings directly into the drive rail.
  

  

• 4” wheels that are a custom 3d printed core with pockets sandwiched by 2 aluminum plates with nitrile tread in between and attached to the 3d print (Credit to Team Tators 2122 for the idea).
  

  

• Custom gearboxes that use 3 Falcon 500 motors (per one gearbox) that are geared to a ratio of 8.8:1 which is achieved by 11t gears (on motors) driving a 50t gear which is on the same shaft as a 28t gear (these 2 are the idler gears) which drives a 54t gear which is the final gear and it drives all 3 wheels on its side (this gives us 8.8:1 ratio, 684rpm, 11.95fps gearboxes).
  

  

• Finally our last feature other than bumper mounts, triangular pockets and a top belly pan, is the thing we are most excited for and it’s the drive rail assemblies being their own quick change field serviceable unit so at competition we can bring spare drive rails so if something breaks we can take the drive rails off, put the spare drive rail assembly on and we are ready to go back out for a next match without a janky drivebase and the broken drive rails can be fixed by the pit crew later.
  

  

[coderkevin]

Weekly update

Just like all of the rest of you, we’ve got a lot going on right now!

Drive base

We’ve been making steady progress cutting out the gearboxes and drive base elements, even as we’re still finalizing the superstructure mounting in our CAD.

Climber

We’ve also gotten an extension arm completed with the spool underneath and 1/8" dyneema line running within it. The axle that holds the spool will also be the pivot for the extending arms to rotate forward to catch the next rung up for a traversal climb. Note that we’re using 1/8" aluminum tubing in 2", 1.5", 1" progression with a design based on the one from The Thrifty Bot this year (shout out to @Ryan_Dognaux for allowing us to copy his homework on this one!) We may try a lighter weight version later, but this is one that should be fairly torsion-proof.

Here’s a video of us running the spool up and back:

Climber extension working

Note that the above video is using 3D prints for the 1/4" plates at the top. We’re planning on cutting those out of aluminum eventually.

CAD

CAD madness continues! We almost have the climber design complete, just a few details left. The turret is mostly complete, and the intake geometry drawings are in full swing. Still a lot of CAD to do. We intend to prototype some critical geometries by cutting some out in particle board or plywood to test before we cut the real metal or polycarbonate parts. Iterative prototyping, right?

Another shooter prototype

We’re starting on another shooter prototype, this time with a 2" top roller. Also we’re going to use 2 Falcons for the main roller, and a single falcon for the top roller. This will get closer to the real thing for us and allow us to ensure we can control our back-spin (especially due to the Cavbot’s thread and graciously sharing their observations on backspin!)

Programming

Our programming efforts have been a mixture of a few things. We’re supporting the prototyping efforts, but also grabbed the latest trajectory code from WPILib to ensure it works ok on our older 2019 drive base we’re using as a test platform and will later use as a defense bot for practice. Also, we’ve got some of our first-years working on PhotonVision for detecting cargo balls, and the Arduino sensor network. We think we’ll be using 2 hall effect sensors (one for turret, one for adjustable hood) for zeroing, and a few IR proximity sensors for the indexing. Other than that, it’s basically just encoders. We will be using our usual Grayhill 256 encoders on the drive base because we really like the 1:1 encoder output, but elsewhere, we’ll use just the encoders on the motors.

At any rate, hope you’re all doing well and keeping busy! Cheers!

[mmongar]

Thanks for sharing System Meltdown. I know that we brought up Open Alliance with our team 5013 and we just didn’t feel like we could commit. We will mention it again next year. See you at Central Missouri.

[coderkevin]

Of course! We’re trying to capitalize on having a team full of first-years by doing all the things we feel we should!

[Bmongar]

6 falcons at 8.8:1 ratio, 684rpm, 11.95fps Planning on pulling a trailer with that much power? Wow, impressive.

[JoshuaP]

1. Why 3D printed wheels instead of, say, a COTS aluminum wheel?

2. Why is the drive gearbox geared for so much torque? I’d argue you can go much faster with 6 Falcons. Also, you might suffer wheel slip with that much torque on the wheels.

[coderkevin]

Yeah, it sounds like a lot! We were pretty surprised when we saw teams running 6 falcons. There are some compelling reasons as far as power curves and amperage consumption for a more efficient drivebase. We ran into some stalling problems with 4 falcons in the past where they “squeal” with quick vector changes, so we figured we’d try out the 6 falcon approach this year.

[coderkevin]

Good questions, thanks!

For the 3D printed wheels, we see a few benefits. First is we can choose any wheel diameter we want. We’re really close to the ground right now and may have some issues with the 5/8" wire channel. If we do, we can increase the wheel diameters. Second is weight. The 3D printed wheels are definitely lighter than some of the other options out there.

When we decided on our drive gearing, our goal was to hit about 12 feet/second at peak efficiency of the motor curve. We believe doing that with 3 falcons per gearbox will help our power usage, minimize stalling, and use less amperage per circuit.

[ngreen]

I’d say cost and turnaround time is also an advantage if they work well. Back of envelope has it less than half the cost of COTS with your own labor and machines. So that would be nice for having more spares, and using on offseason projects, etc. And if you need different wheels tomorrow you don’t have to wait on it being delivered.

[coderkevin]

Hey folks! Sorry it’s been so long since the last update. I’ll catch us up here, but I’ll split the more informational stuff into a few posts below. Here’s the summary since the beginning of last week:

Last week:

• We did a second shooter prototype, this time with a top roller and powered by Falcons
• LOTS of CAD updates
• Research on LiDAR for cargo following
• Several programming updates

And we ran into a couple of issues:

• We tore up an aluminum gear during burn-in of a main gearbox that wasn’t properly fastened
• Our Omio X8 router succumbed to a series of unfortunate events, which set us back on our CNC machining

This week:

• Snow days Wed-Fri, so we have very optional workdays Thursday, Saturday (today) and Sunday (tomorrow)
• Today we finally got to test out our drive base completely. Way later in the schedule than we would like, but we’ll catch up.
• CAD is 95% complete for Rev. 1

[coderkevin]

Revisiting the shooter prototype

A little more than a week ago, we decided to revisit our shooter prototype. This was in response to the several videos we’ve seen of bounce-out being a real problem for scoring in the upper hub. (Thanks @CAVBOTS)

So, we stepped up our prototype a bit, and used 3 Falcon 500 motors. 2 on a WCProducts 4" roller, and 1 on a set of 4 2" stealth wheels.

We tried several iterations mostly running in the 2000 RPM to 4000 RPM range on both top and bottom rollers.

This is the “control”, with no motion in the top roller, and the bottom roller at 4k RPM

Shooter Prototype v2, stationary top roller

This everything the same, but with the bottom roller at 4k RPM, and the top roller 4k RPM

Shooter Prototype v2, reduced backspin

Quite a bit more power and distance, and of course, much less backspin.

We tried enough other speeds and launching angles to be confident we can manipulate the distance of shot and backspin sufficiently.

Summary:

1. We definitely want a top roller for our shooter this year.

2. We believe we can control our distance and angle enough with the top/bottom roller speeds, so we will build a shooter with a static hood angle at 30 degrees (the longest angle we need, as it’s easier to adjust speeds down for shorter shots)

3. 4" main roller and 2" top roller seem to work well. We tried with compressions ranging from 1" to 1.5" and we’re settling on 1.25" because it seems to give us better predictability with deflated cargo (down to about 2.5 PSI)