Celt-X 5406 | 2024 Build Thread | Open Alliance

[nuclearnerd]

Celt-X 5406 is the robot sports team from Bishop Ryan Catholic Secondary School in Hamilton, Ontario, Canada. 2024 is Celt-X’s 10th year of competition, and our first year joining the Open Alliance. We’ve always tried to share our resources and experiences with other teams, but this is the first time we’re going to try to document our build in (somewhat) real time. We believe in the OA mission, and we hope this thread helps to foster a culture of collaboration and innovation within FIRST!

2024 Kick-Off Plans

Celt-X meets Tuesday, Wednesday and Friday evenings, as well as Saturday afternoons, but this kickoff weekend we're throwing a Sunday in as well. Over the course of the next two days, we hope to:

• Understand the new game
• Set goals for our new season
• Figure out what design and performance requirements are needed to achieve those goals
• Map out a plan for meeting those requirements

To that end, we’ll be following a kick-off worksheet which we’ve been tweaking for many years. I’ve attached a copy below as our very first OA contribution. Feel free to copy and modify it for your own team!
2024 Celt-X KickOff Guide.pdf (99.9 KB)

2024 Build Schedule

Starting Tuesday we'll jump into prototyping and designing our robot. Last year we did a really good job of analyzing the game, but we struggled to achieve the lofty performance goals we set for ourselves. So In 2024 we're going to try something new. Rather than trying to build the "Einstein Robot" in one go, we're going to take a more iterative and incremental approach, taking a page from teams like 1678 and 3847. If all goes well we'll build 3 robots before our first event:

1. “ZeroBot” will be barely more than our 2023 summer robot chassis with some superstructure prototypes strapped to the top. This robot will never compete, but will give our programmers early access to a driving swerve chassis for configuring the 2024 software, and will give our build team a platform to test their prototypes. If all goes well, this robot will move its first game piece by January 20.

2. “AlphaBot” will be designed for competition, but will be limited to the capabilities that we can build quickly. The goal will be to celebrate Alpha’s “Frankenstein Day” (when the robot first comes alive) by the end of January, when the students break for exams. The design will be as modular as possible so that we can swap out mechanisms as we find time to improve them.

3. “BetaBot” will hopefully share many of the same modules but with improvements. We won’t start the design until after exams, when we’ve had a chance to learn from Alpha. The challenge will be to complete the design and build of Beta by the end of February, so that we have a couple of weeks (including March Break) to program it and practice it before our first event on March 16.

If we do it right, we’ll have a driving, competition robot available on the practice field for programming and practicing from February 1st onwards. If Beta (or subsequent) revisions don’t work out, we’ll always be able to fall back to Alpha.

Here’s how that schedule looks as a Gantt Chart (using https://www.onlinegantt.com/).

• Blue bars are tasks for the design subteam,
• pink is for the build team,
• green is for the field team, and
• purple is for the programming team.

Since this is the first time we’re attempting this plan, I expecting that the time estimates aren’t perfect. The power of a Gantt chart is that we we can see what the consequences of any slips are on the whole project schedule. We will try to keep this chart updated weekly as we go, and I’ll be ruthless in cutting scope so that we don’t get caught behind schedule again. Wish us luck!

File that can be opened in OnlineGantt: CX24 Build Schedule 20240105.gantt.txt (11.4 KB)

As well as schedule updates, we’ll be asking our student leads to use this thread to post some of their successes, failures, and lessons learned as we stumble through another build season. We’ll try to update the thread weekly, but I don’t want to over promise. In the meantime, if you have any questions or feedback feel free to post them here - we’re happy to hear it!

Good luck to all teams in 2024. We hope you have a fun kick off!

[marshall]

Looking forward to this one too!

[soderkk]

Saturday (Kickoff):

For Kickoff we met at our school to all watch the kickoff video as a Celt-X community. We all piled into the auditorium in anticipation, and shared the excitement when Cresecendo was finally revealed. Afterwards, we rewatched the game animation a few times, then all shared a potluck lunch before splitting into groups to dive into the game manual and break it down.

We spent a couple hours breaking down all the field elements and rules, trying to understand how everything worked. Once each group had read through it all and had taken some notes, we met back up as a team and delved into the game further, trying to answer any unanswered questions. Here’s our end result:

2024 Celt-X Game Questionnaire Answers.pdf (223.6 KB)

It’s worth highlighting the questions we couldn’t answer on Saturday:

• 6.5.1 Trap scoring - how far past the trap door counts as “entering”
• 6.5.2 Seems to conflict with other climbing rules. Does “directly supported by chain” for Harmony mean fully or only supported by the chain (not other parts of the Stage?). We’re having nightmares of our 2019 Einstein finals!
• G424 - does jumping count as a double tech foul if you get hit? Can you climb either alliance stage, and if so how do those fouls work then?
• G408 can robots score high notes on microphone? Any penalties for using A-Stop strategically?
• Are TPU wheel studs (getting popular) illegal per R201?
• G428/9 how far can human player make notes go? (need to test on our practice field)

Sunday (Strategy Day):

Goals:

The first thing we did was decide on some goals as a team for this season. Here’s the list of all the ideas we had and what we decided on:

Competition goals:

• Win a blue banner (specifically an event win)
• Qualify for Worlds for realsies (Seed high enough to qualify without other teams saying no)
• Seed top 16 in our division at Worlds (Captains/First pick)

Other performance goals:

• Build 3 robots (0, α, β)
• No on-field breakdowns (including not requiring the A-stop)

Culture goals:

• Run FIRST Aid event again, an event we last held in 2020 where we help teams who are struggling get their robot working before their first event.
• Do no harm in service of competitive goals (monitor grades, pay attention to mental health)
• Cut scope to prevent “midnight miracles” (Staying up until ungodly hours to get things done when we’re running out of time)
• Clock at least 2000 cumulative Student hours by end of February (which would be just a bit over 50% attendance)
• Improve our online engagement. All members will check our Basecamp site/app at least once per day, and we’ll update our Open Alliance thread at least once per week.

We think these are ambitious, but achievable, and we will use them as the basis for most of our planning and design decisions.

Trying Out XRC:

Afterwards we took the time to try out the XRC version of Cresendo, trying to get a feel for how everything worked and get a rough estimate for the amount of points a robot can score.

Here’s what we learned:

Without any strategy:

Robots averaged a score of 15-25 cycle points in a 3v3 match. This was without amplifying (we found it nearly impossible to do with the current robot build in the game) and without any defense.

With defense:

Robots averaged a score of 0-20 cycle points in a 3v3 match, with a robot on each alliance playing man on man defense. However, this resulted in ~50 points in tech fouls. We found that although powerful, defense in this game must be incredibly careful and is really susceptible to fouls from the podium and amp zone. We also found that defense was really good against robots that had to score right up against the subwoofer, and could shut them down completely, which makes us think that good defense will be especially powerful this year.

Skills and Strategies Brainstorming:

Next we spent some time trying to figure out all the different things a robot might do this year and how difficult each task would be to implement. We also sorted them into categories of what we’d like each robot iteration to have. Here’s a copy of the spreadsheet:

2024 Robot Skills List

Other than tasks, we also wanted to figure out all the different strategies a robot might have in a match. We did this using stickies.io, which we find is a great way to brainstorm collaboratively.

Here’s what we came up with for:

Auto:

Teleop:

“Endgame”/ Stage:

We then estimated what different “tiers” or robots would play which strategies, and how well they’d score on their own. Note, to simplify the estimates below, we considered each Speaker cycle as 2 points, and each robot only gets credit for its own climb points. Of course amplified cycles would improve these numbers.

BLT (Brave Little Toaster):

Leave: 2
Score preload: 5
2-4 Teleop cycles: 4-8
Park: 1

Total 12-16 pts

Average Ontario Robot:

Leave / 2 piece: 12
4-7 Teleop Cycles: 8-14
Climb: 3
½ of Harmony: 1

Total 24-30 pts

Average Houston Robot:

Leave / 3 piece: 17
6-12 Teleop Cycles: 12-24
Climb + Harmony: 4
Some Trap: 0-5
Spotlight: 1

Total 34-51 pts

Einstein Level Robot:

Leave / 4-5 Piece: 22-27
10-20 Teleop Cycles: 20-40
Climb + Harmony: 4
Trap: 5

Total 51-76 pts

Setting Design and Performance Requirements

We can use these estimates to try and guesstimate what robot we would be required to build in order to be competitive enough to achieve our goal of winning an event (in this case our first event is Durham):

What’s Required for Winning Durham:

Playing finals against 2x Avg Houston + BLT = 80-120 pts
Us + one BLT + one Avg Ontario = Us + 40-45
Therefore, to win that match, Us = 40-75 pts on our own
(OR guarantee we rank 1st or 2nd).

To achieve this target, we laid out a roadmap for our three planned robots:

ZeroBot Requirements:

• Running by January 20 (!)
• Mount the kitbot shooter on Q-Bert
• Equivalent to a BLT (6 cycles)
• Nice to haves: Climb, Basic Vision

AlphaBot Requirements:

• Running by February 1 (!)
• Fast, swerve drive base
• Modular superstructure for easy upgrades
• Short enough (25”) to fit under the Stage
• 2 to 3 piece auto
• Minimum 10 cycles in Teleop
• Score in the Speaker from anywhere in the Wing
• Score in the Amp
• Floor intake. Must be fast, indestructible and foul-proof. Might be over or under bumper.
• Single robot climber
• Vision alignment with April Tags
• Nice to haves: Can shoot the High Note onto the Microphone, Shoot on opposite side as intake (pass through), unblockable shot (high release if needed)

BetaBot Requirements:

• All the requirements of Alpha, plus
• Running by March 1
• 3 to 4+ piece auto (including the centerline notes)
• Score 10-15 cycles in teleop, with headroom to improve for later comps.
• EITHER score in the trap or 100% reliable at scoring the High Note (lock in the Ensemble RP)
• Nice to haves: Shoot while driving, global field localization with April Tags, Cheesecake climber

The targets have been set, and they’re ambitious! Hopefully we can hit them!

Human Matches and Testing:

We also spent some time playing around on a mock field, playing some human games and trying our hand at scoring the highnote. For reference, out of about 100 attempts, only the one shot below, was successful

The human matches were a lot of fun, and we had some people play defense, or different types of robots (BLTs moved at a walking pace, Houston robots could speedwalk). We found that the human game matched our estimates from earlier pretty well. 15-25 alliance cycles undefended, but that was cut down by half with “defense”.

Lastly, we broke out our off season robot from last season, Q-Bert, to see if it had potential to pickup and shoot a note without modification. We found that although it couldn’t intake the note, if we manually fed it, it could lauch the note a few feet. Definitely not far enough or high enough for the speaker, but it’s a good first step! Next week, we’ll strip the 2023 intake off Q-Bert, and replace it with a kitbot shooter. This will become “ZeroBot”, which we’ll use to gather early on-field Crescendo experience.

Note that we really didn’t do any design in the first two days. That was intentional! We don’t like to start solving problems until we understand them. After two days of analysis, we’re finally in a good spot to start research and development over the next few weeks.

Overall, it was an intense but fun weekend. We can’t be more excited to see how this season unfolds!

[nuclearnerd]

5406 Open Alliance Posts for Jan 9 to 14 2024

Celt-X finished our first week of Build Season on Saturday. We got a lot done, made some conceptual design decisions, and learned some lessons we can share. As requested, I’ll break this week’s update into a couple of separate posts…

[nuclearnerd]

Prototyping

The team Identified that the robot will need 7 systems or mechanisms:

• Chassis/Drivetrain,
• Intake,
• Feeder (maybe),
• Speaker Shooter,
• Amp Scorer (likely shared with Trap?),
• Climber, and
• Camera Vision.

We took that list and identified the unknowns for each system that might require prototyping to answer:

With the “unknowns list” complete, we set about building prototypes to answer them.

Drain Pipe Amp/Trap Mech

On Tuesday, rule update 1 clarified that robots could not shoot the high note. So that means that our Beta robot will have to score in the Trap according to our design requirements. Knowing that we won’t be able to reach much higher than the bottom of the Trap, we decided to try shooting the note downward using a “drain pipe” mechanism, along similar lines as other OA teams.

A unique feature of our prototype was to see whether we could squash the Note into a smaller width (8”) at the same time it went around the hook. If so, it would allow more room to be misaligned when scoring in the Amp / Trap.

Our first attempt with a single roller and a fixed (lexan) hood didn’t work. It had too much friction - the wheel just rubbed away the Note surface

We made a second attempt, this time replacing the plastic hood with unpowered idler rollers and adding some low-friction tape to the side plates. This worked really well! We will likely incorporate something like this in our robot.

Zerobot

We removed the intake from summer robot “Q-Bert”. Then we put together a kitbot shooter, with minor modifications to the shooter so that it could attach to Q-Bert’s wrist.

Presenting Zerobot! It has no intake, and it isn’t shooting yet (needs more code this week), but it’s pretty good for two days’ work. This will give us some good early data about shooting performance, and cycle time. On-field testing will start next week.

We also flipped the swerve modules to increase ground clearance so we could test driving over a gamepiece. Preliminary results show that the 4" wheels have no trouble getting in and out of gamepieces, but we’ll experiment more this week. We’re not sure whether it’s better to drive over gamepieces with high bumpers, or push gamepieces away with low bumpers (and risk getting bounced on top of one).

Ongoing prototyping

The schedule has a few days allotted for more prototypes. Here's what we still want to do:

• We need to figure out if there is a shooter design that outperforms the kitbot. To that end, we started work on horizontal roller shooter prototype using laser cut plates on Friday. We’ll hopefully have results as early as tomorrow.
• We need to figure out how to intake under bumper. We started work on intake geometry prototype using Thriftybot 3DP clamps and HypeBlocks on Friday. But we’re encouraged by the successes other OA teams like 4522 have had with “in-bumper” intakes.
• We have some ideas about climbing geometry, but we’re 2aiting on field completion to test climber prototypes.
• On field testing of zerobot starts this week!

[nuclearnerd]

Design and CAD

On Monday we split off a small Design and CAD team to prepare some “Crayola CAD” concepts.

Climber
Given that climbing is one of the hardest tasks, and will require a lot of space on the robot, we looked at the climber geometry in Onshape.

• The climber needs hooks that can raise up to ~4 ft (to climb on the sides of the chain), and lower to ~1 ft (to get the Trap mech up to the Trap).
• The robot CG should be between the hooks and the Stage Core so that it tends to tip towards the Core (after the bracing tower is extended)
• The climber needs a bracing tower with wheels that can roll up the wall of the Stage Core and support the robot from tipping when climbing. The chain will angle back to counter the pushing force from the tower, but they should balance without too much angle.
• The top of tower should carry the Trap mechanism to max height. Even then, we only get a few inches above the bottom of the trap door, so trap mech needs to be small.
• Ideally the hooks face towards the Stage to catch the chain as we drive in so we don’t have to place ourselves perfectly between the chain and the Core before climbing. The challenge is that chain can swing towards the core, leaving very little room to deploy the bracing tower.
• Another challenge is that chain needs to be pulled down as low as possible (right to the bumper), so the chain “path” has to cut right through the middle of the superstructure.
• A solution might be to have the CG location (relative to the hooks) change as the robot climbs so that the hooks pull the chain around the superstructure, and the robot tilts away from the core until the trap mech and tower are deployed, and then tilts back towards the Core. We’re still working on how that might work.

Gamepiece Manipulators (Shooter, Intake and Trap Mech)

Next we looked at our options for manipulating the game pieces. Decided there were at least two choices to make:

• We have a requirement to intake gamepieces from the floor. But do we do that over the bumper, or under the bumper?
• We have a requirement to score in the Amp and Trap. What mechanism do we use for that? The shooter, the intake, or a dedicated mechanism?

Those choices gave us a grid of 6 potential robot architectures. We looked at what a robot might look like for each cell in the grid, what advantages and challenges each might have.

	Under Bumper Intake	Over Bumper Intake
Intake does Trap	Impractical. Robot would have to rotate to present bottom of bumper to Amp/Trap	Intake needs to lift to Trap height, as well as reaching over bumper so it needs 2 degrees of freedom
Shooter does Trap	Shooter needs a “cowl” to redirect the shot downward when scoring amp/trap.	Shooter needs 2DOF mechanism to lift from 2ft under-stage limit to 4’ trap height
Separate Trap Mech	Need two handoffs between intake and amp/trap mech. Could be a feeder switch, or shooter does the passing. Trap mech needs to elevate from 2’ tall max to 4’ tall max.	Same as left, but with a separate intake that pivots over the bumper. Intake could have separate handoff positions for each mech. Same intake disadvantages as above.

The advantages of under bumper intakes are they’re faster to deploy, more damage proof in a collision, and might be able to intake from more than one direction. The disadvantages are that it is less “modular” for future upgrades, and may not have as wide a catch area if the game piece has to pass between the swerve wheels. On the whole though, the team decided the advantages outweighed the disadvantages. That left us with only two preferred architectures, which greatly reduces our solution space:

1. Under Bumper Intake, Separate Trap Mech
2. Under Bumper Intake, Shooter does Trap.

With that choice made, we split the CAD team into two groups to mock up our two preferred architectures using KrayonCAD (which is awesome btw @Nick.kremer). The sketches below aren’t anywhere near to final designs, they’re just a first attempt to communicate what each robot might look like, to help us identify any particular advantages or challenges. Each subsystem might end up looking completely different to what’s shown.

At the end of Friday’s meeting we conducted a “Conceptual Design Review” (CDR). Both design groups presented their concept to the full Celt-X team. This was a super fun moment, which took significant presentation and listening skill. We’re very proud of these students!

CX24 Robot Concept 1 Presentation.pdf (455.1 KB)

CX24 Robot Concept 2 Presentation.pdf (641.1 KB)

The team asked a lot of good questions, and they identified advantages and challenges of each concept:

	Advantages	Challenges
1) Separate Shooter and Amp/Trap Mech	More modular, with separate mechanisms for separate jobs. More stable shooter (since it doesn’t lift)	Climber needs 2DOF to move CG during climb. Extra mech might be heavier. Requires a feeder “switch” or a double-handoff. Handoff must be reversible so drivers can score either location quickly.
2) Shooter does Trap	Might only need intake to shooter handoff. Shooter lifts out of the way of the climber hooks, can be used to shift CG during climb. Shooter lift axis could help take shots from a less defendable angle. Could be a candidate for “creativity award”.	More complex, heavier multi-function shooter Less stable base for accurate shots (Could be mitigated by having the lift axis be latched most of the match, and released for the climb only, but that would be a risky “one shot” mechanism.)

Both concepts appear to have the potential to meet all of our design requirements, which is nice! The team identified several challenges that are common for both concepts:

• How will the under bumper intake package? Can it intake from the opposite end of the shooter for a pass-through? Can it intake from both ends maybe? Can it intake from all four sides maybe?
• Will the mechanisms interfere with each other once we give them more reasonable dimensions? Will the CG be balanced, or all on one side?
• How will intake interact with Swerve modules - can gamepiece get trapped under the robot?
• Where will the cameras mount? We need two cameras - one for april tags (looking in scoring direction at ~5ft elevation), and one for gamepieces (looking in intake direction at floor in front of bumper). Those will look in opposite directions for a pass-through shooter.
• Where will electronics mount (especially if under-bumper intake takes the traditions 5406 location)?
• What is the chassis size and design? We need to finalize that immediately to keep on AlphaBot schedule. That’s tough to do right now without all the answers for how the under-bumper intake and handoff will work .

We’ll try to answer those questions, and choose a preferred concept to develop further in the next meeting or two. Stay tuned!

[nuclearnerd]

Field Build

The field build team dove head first into preparing our practice field. On Tuesday, Wednesday and Friday they:

• tore down the remaining Charged Up field elements
• patched the carpet
• repositioned the driver stations and sidewalls to match the Crescendo field
• Modified two field wall segments to form the angle by the Sources. Whenever we build feeder stations in plywood, they get destroyed during practice. These new borders are made of lexan and aluminum and rigidly connected to the rest of the field, so they should be much more durable!
• Modified the “Team Version” plywood element designs to suit our (nearly) full size field. These include completely new designs for the Source and the Stage. You can find CAD here.

On Saturday we held our annual “Field Build Party”. 40-some students, mentors and parents from teams 5406, 4039 and 9062 got together to build a full set of field elements from wood.

Incredibly, this is the first year that we’ve (almost) completely finished the field in a single day! We’re really grateful for all of the hard work from everyone who attended Saturday!

Some teams have been asking about practice field sign up. We’ll be opening the field for up to two visiting teams each Saturday. Details about the signup process will be posted soon!

[nuclearnerd]

FIRST Q&A Questions

On Thursday, the FRC Rule Q&A system opened for questions. Celt-X submitted two of questions, which resulted in some important clarifications and revisions:

• The Stage Core can safely be contacted during an “Ensemble” climb
• Robots can only climb their own Stage for OnStage points (and fouls)
• We also learned from other’s questions that
• Robots can no longer score Highnotes.
• The Trap Door is now one of the legal contact surfaces for climbing.

We’ll monitor the Q&A and ask additional questions if our other kick-off uncertainties aren’t asked and answered soon.

Schedule Tracking

An updated version of our project schedule Gantt Chart is attached below.

CX24 Build Schedule 20240114 (4).pdf (144.6 KB)

• We were on schedule for ZeroBot, and ahead of schedule for Field Build, which is awesome!
• Unfortunately our conceptual design and investigative prototypes are falling behind our (too aggressive) schedule. So we didn’t get to start designing AlphaBot this week.
• As a result, the schedule shows AlphaBot completion has slipped to the second week of February instead of Feb. 1. That will make it a bit harder to get a second iteration designed and built in time for our first competition, but it’s not time to cancel BetaBot just yet. We’ll monitor the AlphaBot progress and decide on Beta in a couple of weeks.
• Right now, the critical path is getting the AlphaBot Chassis designed and constructed. That means we have to make some decisions about frame size and intake right away. The pressure is on!

If you’ve read this far, we humbly thank you for your attention! Please feel free to ask any questions or offer us any advice. Onward to Week 2!

[nuclearnerd]

Quick update. We finally got ZeroBot running cycles tonight. It can intake from the Source, and shoot from the Podium or Subwoofer.

Video here

Oh, and here’s a nice surprise, ZeroBot can also score in the trap! Thanks to SuperNurds for the inspiration!

Video here

Slow motion

[T3_1565]

How accurate do you think the team built trap is compared to the real thing? It looks repeatable from a robot perspective.

Seeing this success would reduce a lot of complications in our bot if its reliable!

[nuclearnerd]

This trap is built to the team element drawings. You can compare the drawings / 3D models between the real field and the team elements. The geometry and surfaces are pretty comparable though. We have a lexan door that isn’t yet hung, but otherwise I’m not sure what to change to make our trap “more realistic”.

[T3_1565]

So your opinion is its very likely to also work on a real field?

Thats good to know. Thanks for the video!

[HUGODANIEL09]

@nuclearnerd

Oh, and here’s a nice surprise, ZeroBot can also score in the trap! Thanks to SuperNurds for the inspiration!

Hi, that’s awesome!

Is it consistent to do it in match?

[nuclearnerd]

@HUGODANIEL09

Is it consistent to do it in match?

No idea yet! But with the “Zerobot” parked in place, we could make the shot pretty much every time - so it would just be a matter of getting to the same place (plus or minus some amount we don’t know yet).

[Skyehawk]

Any insight you could share in how far away from the source you are? (Or the size of the robot frame? )

[nuclearnerd]

Robot frame is 28 x 24. We didn’t take an exact measurement of distance, so your estimate is as good as mine. (The shot was lined up by eye, as a last minute test at the end of yesterday’s meeting).

[Skyehawk]

Seeing you all managing a shot (great work btw) is leaving me less and less worried about any special mech for trap or even bothering to make the geometry close-ish. Get far enough ahead in the match score, shoot at the trap 2-3 times from a known spot (think touching the pyramid with zip ties in 2013) and Bob’s your uncle. RP on ez mode when in a pinch.

[yesiconcur]

Isn’t there a flap in front of the trap? how are you supposed to shoot through it?

[nuclearnerd]

@yesiconcur

Isn’t there a flap in front of the trap

Look closely at the video. The flap is there.

[DOOF]

The trap shoot seems to be a reliable and honestly very viable option, I am just worried (and frankly hopeful) that FIRST fixes it in the next team update (if they’re thinking of making an update it needs to be ASAP). Only time will tell but this isn’t a 4907 jump moment where only one team will have it so it’s not broken just a fun thing it’s not something that was planned and lined up by eye in like 20 minutes and was consistent… scary (Celt-x alum that was driving the bot in this clip for cred lol)