FRC 4272 Maverick Robotics - 2024 Build Thread

[Lexeo]

About us

Welcome to the 2024 updates thread for team 4272 - Maverick Robotics! We are based out of McCutcheon High School in Lafayette, IN.

We are proud members of the FIN district and will be attending the Mishawaka (week 1) and Columbus (week 3) events.

We will post general updates, successes, failures and progress updates. Our goal is to highlight the great work that we are doing. Follow along with our Github and Onshape for live updates. We welcome any feedback and hope other teams can learn from our successes and failures.

We are excited to participate in openalliance for the first time this season!

Important links

Team Website

Facebook

GitHub

2024 Robot (Onshape)

[Lexeo]

Our preseason

Since May, we have been working to train our team, and to explore new techniques and designs that we can use during season. Posts over the next few weeks will highlight some of those projects that other teams may find useful…

Spark Max Accessories

We spend a lot of design time trying to optimize and clean up our wiring on our robot. For the 2023 season we designed a custom Spark MAX mount. Overall, we liked this because it helped keep our signal and CAN wires connected.

For 2024 we replaced the zipties with bolt-on 3D printed wire clamps. This makes it easier to service and hopefully gives us an even more robust connection.

Our Onshape model for this part also has more configurable options, to have more options for mounting.

Inspired by Team 3171’s Hurrilink and Hurrilink Sidecar, we decided that mounting the Spark MAX to the Neo is frequently a great solution. We wanted to combine our Spark MAX mounts with the Hurrilink concept to create a solution we could use for 2024. We designed both a top mount and a side mount. Overall this seems to be a really nice way to mount a motor controller to organize a lot of the wires.

To take this concept to another level, we started looking at how our Spark MAX’s could attach to our MK4i Swerve modules to make a clean and contained solution. This is what we came up with. . This reduces the overall footprint of the module, hides the wires, and makes the system more modular.

[Lexeo]

Happy New Year everyone!! Wanted to share some additional projects we worked on this preseason

PCB Designs

Back in the 2023 preseason, 4272 released some PCB designs to help us simplify our wiring and help standardize to Molex SL connectors for our small gauge wires. Full write-up of those designs can be found here. All of the 2023 designs were used on our 2023 robot and worked well for us.

To do this work, we are using KiCad EDA. This is a nice tool and is free to use. It has been a great tool to teach PCB designs, and it’s even something students can easily use in college and beyond.

For 2024, we had a student take advantage of the Hack Club grant to manufacture the prototypes he worked on. He even got the project showcased on the FIRST community blog. If other students are interested, check out Hack Club OnBoard.

MAVcoder R4

This is our 2024 version of our absolute encoder.

Like the earlier versions, this is based on the AS5600 IC chip. By default, our encoder is designed to be powered at 3.3V, and will output an analog signal from 0-3.3V depending on the angle of a nearby Diametrical Magnet.

This new board is significantly larger than the previous version and bolts to the swerve module with 8-32 bolts - replacing the aluminum cover on the MK4i. This extra space gave us room for a few new features:

• Rather than soldering wires directly to the board, we are instead using a Molex SL connector. This makes the wire easy to replace if needed. This also makes the board assembly easier.
• We were able to break out the i2C pins on the board to a header. This allows us to configure the IC using this kit. Our plan is to configure them to PWM output, which is a little less susceptible to noise. This is also in response to some strange behavior on a few SPARK MAXs where the analog input voltage would read up to 3.7V and be more noisy. This happened at an off-season competition last season and required us to replace the SPARK MAX to work correctly.
• Using the programmer, we also believe that the wheel offset can be adjusted. Each encoder can be paired with a swerve module, and then we can avoid changing the offsets in the robot code.

Full project design files are here for anyone interested

MAVprox

This board is our newly designed magnetic limit switch. There are several magnetic limit switches on the market, but none of them had the Molex SL connector we prefer to use - so we designed our own. This board plugs into our SPARK MAX breakout - and should also work with the SPARK Flex.

We designed two versions of the board. One has a through-hole Hall effect sensor, and the other has a surface mount hall effect sensor. This gives us different options on how we position the magnet to the board.

Through Hole Design

Surface Mount Design

MAVterm

Not much to this board. We wanted a clean solution to putting a terminating resistor on our CAN network with a Molex SL connector. We also added a hole so it could be mounted.

Design files

[Lexeo]

We have a few more misc projects to share before kickoff. These are some small helpful things that we’ve experimented with.

Inverted electrical panel

In an effort to try to make serviceability on our robot easier, we designed an inverted belly pan for a chassis, that allows for all electronics to be accessed underneath the robot. Overall, this seems like a good concept for certain games- especially those with a flat field. This is something we may consider using in 2024.

Onshape CAD can be found here

3D printed rack and pinion

As a team, we’ve been thinking about alternatives to pneumatics.
2023 was our first robot to avoid a pneumatic system since 2016. We found this very convenient since a pneumatic system adds weight, takes up space, and adds more failure points.

To help with this goal, we designed a simple 3D printed rack and pinion that can be used with a CIM-style motor. This concept seems promising for light-duty applications - similar to what small diameter cylinders can do.

Onshape CAD can be found here

4-bar intake

We didn’t get around to fabricating this one, but we worked on a design for a 4-bar retractable intake for the 2022 game - avoiding pneumatics. A bit of refinement would be needed to make this concept robust, but we learned a lot making the concept.

Onshape CAD can be found here

[Eric_Sells]

Kickoff Event Structure

For the 2024 season, we decided to run our Kickoff event slightly differently than past years. We wanted to slow the decision process way down and make sure we had full coverage knowledge of how the game actually works before making any decisions that could make or break the season. After watching the game reveal, most of our mentors separated from our students. The mentors took this time to talk freely with one another and to started the field build. The students then had a few activities to get a better handle on the rules. We took them through a worksheet to get them to read through everything. Then we dove into how robots might play the game, but not necessarily how our robot will play the game. The goal of this was just get their minds churning and get the gears running. While this can definitely be a bit more boring than coming up with a ton of designs, we had quite a few students praise it because it felt a lot more lax and was nicer to get a full handle on the game.

Field Build

On Sunday, we had a large group of mentors come in and spend all day building the field so that we could use it on Monday. We usually do not have the field ready for about a week, but we really wanted it done for Monday. We appreciate our mentors greatly.

Human Robot Strategy Testing

Our first meeting, post kickoff, was Monday. Our plan for this meeting was to review the progress of Ri3D teams and to see what those groups were working on. This really helped new students who really had no idea what is going on, but also helped introduce new ideas to veteran students to help them launch even more ideas. After this, we played “Human Crescendo” on our newly built field. We had three team members act as robots, gave them a strategy, and ran the game in time. The first few runs were not as structured, to give the team an idea of spacing on the field. Then we introduced a defensive human robot to the mix and nailed down some test strategies to play out. After each match, we tallied the scores to see which strategies worked better than others. The goal of this was to play out strategies in real match time and to also get a feel of how fast everything might go.

Our observations from this activity are as follows:

• Having all 3 robots actively making sure the AMP is always ready to be charged increased points quite a bit. Essentially as soon as the amplification ends, the next two notes should be immediately scored in the AMP. This is in contrast to having one robot focus on AMP while the other two focus on SPEAKER. (With Team Update 01, this is even simpler because we know exactly how many we can score in the SPEAKER before the amplification ends.)
• Having notes on the ground at the SOURCE when robots got there decreased cycle times quite a bit as opposed to having the HP feed once the robot got there.
• A defensive robot sitting at the opposing alliance’s SPEAKER is much more detrimental to scoring than originally thought by our team. This is something we actively need to be thinking about when we think about strategy and design.
• Assuming all three robots have a ground pickup, getting 6+ notes scored in AUTO seems trivial. Even at a slow pace, scoring the 3 pre-loaded and the 3 in the WING happened every time. Obviously during qualification rounds, not everyone will have a shooter or ground pick up, but assuming one bot could get 3 or more, those 6 score every time.
• Based on our speeds, the MELODY rank point is also trivial, assuming all three robots can shoot. Again, there will be some qualification matches where this assumption is false, but the MELODY rank point should be happening most of the time.
• In contrast, the ENSEMBLE rank point is much harder to achieve. Without 3 robots climbing and a HARMONY, the alliance must either trap, spotlight, or both. This was thought about before the human testing, but became even clearer after.
• Spotlighting is not as hard as previous HP scoring games, but is still not something that should be counted on, therefore scoring in the trap is almost required for qualification matches.

[Eric_Sells]

Robot NEED vs NICE vs NAH Lists

After our human testing, we found quite a few strategies that worked and had some realizations. The following meeting, we watched more Ri3D videos to help with design ideas, then dug into strategy. We made a list of things that the robot NEEDS. These are things that we NEED to do to be competitive this year, in our opinion. Without a single one of these things, we will not be as competitive as we wish to be. Then we have a list of NICE things. These are items that would be great if we can get to them or find ways to do them, but if we cannot, the season will turn out fine. Then we have our NAH list. These are things that we do not want to pursue and now that they are on our list, we are done talking about them. Those lists are summarized below with items in no particular order:

NEED

• Ground pick up
• Shooting into the speaker from somewhere
• Being able to score in the amp
• Climbing
• Having a multi-shot auto
• Swerve drive
• Having a “touch it - own it” intake
• Being consistent
• LED feedback for game piece in intake detection

NICE

• Trap scoring
• Short robot - to travel under the stage
• Climb width small - to allow easier harmony
• Long range speaker shooting
• Low weight/high speed
• Low center of mass
• Having the robot know if a note is missing in auto so it can move to the next note
• Automated climb
• Limited game piece movement inside the robot

NAH

• Human Fed Intake
• Buddy Climb
• Pneumatics

“Every Second Strategy”

Using these lists, we then formed our “Every Second Strategy.” This is essentially what we plan to do every second throughout the match. Once this was decided and agreed upon, it is now our strategy moving forward and any changes to it must be discussed by the full team. That strategy is listed below:

AUTO

• Score preloaded piece in the speaker
• Run auto that scores multiple game pieces in the speaker
  • This will vary depending on partners
  • Options are front 3, only mid pieces, or some combination

TELE

• We will run a similar cycle throughout the match with the goal that the following cycle is under 10 seconds
  • Travel to source in the fastest path possible
    • While on the way, the HP will feed a piece to the ground
  • Ground pick up the note in the source zone
  • While returning to our wing, we check the amp
    • If it is not charged, we score in the amp.
    • If amplification is in progress, we shoot in the speaker from as far away as reliable at the given time.
    • If it is charged but amplification is not in progress, we wait for the amplification to start then shoot in the speaker
• Repeat this cycle until T=?? (We will find this time through testing)
• Then we pick up a note if needed
• Head to best stage chain (spotlit chain, or harmony chain, etc)
• Climb to score in the trap
• Score in the trap - stay climbed
• End of match

Using all of this now, we can start prototyping for what our needs are.

[sgeckler]

Pneumatics - NAH! Love it!

[UnofficialForth]

Defense Deep Dive (with Croutons!)

One thing that we decided to do early on in our analysis was a deep dive into defense for this year. This is one of my favorite topics of all time, but it’s been a little bit since I’ve talked about it with others, so I decided to start the discussion with a friend over lunch. The following chaos ensued:

How to Play Defense on a Salad

Defense is often under-analyzed as it is lumped into a single category. We instead broke defense out into 10 different categories, each with varying strengths and weaknesses for each game.

Ten Types of Defense

• Pushing - Forcing an opponent into a pushing match by driving head-to-head in the opposite direction as them.
• Hitting - Hitting an opponent repeatedly while they’re trying to aim for a target so they can’t line up.
• High-speed ramming - Hitting an opponent as hard as possible by driving quickly into them. As opposed to pushing which relies on torque, this method relies on speed to increase a team’s max kinetic energy. This disturbs the flow of the game for the opponent.
• T-Boning/friction Pin - Pushing into the side of an opponent’s bumper, which causes them to spin around your robot.
• Robot Positional - Determine where one’s opponent is and where they want to be, then get in between that.
• Game Piece Positional - Determining where one’s opponent is and where they want to score game pieces, then blocking that trajectory, often with a polycarb shield or t-shirt.
• Crowding - Driving a robot in a high-traffic area makes it hard for robots to drive around. Unlike Robot Positional, often the teammates of the opponent do just as much harm as the defender.
• Clipping - Driving by an opponent at high speeds and hitting the corner of their bumper, causing them to jerk suddenly in one direction. As opposed to high-speed ramming, the goal is not to impart all your kinetic energy into the opposing robot, but rather to be able to do a drive-by and continue a cycle. Can be played well as an offensive robot.
• Game Piece Starvation - The act of taking game pieces from a strategic location of the field first to force opponents to do longer cycles to get their game pieces. This includes holding opponent game pieces or moving them to hard-to-see locations.
• Autonomous Disruption - Disrupting a robot’s predefined actions to result in a less desirable outcome, such as replacing an opponent’s game piece with your own or stealing game pieces they were planning to pick up.

Analysis

We then analyzed the three types of defense we felt would be most prevalent. We agreed upon was that clipping, game piece positional, and hitting would the most common forms of defense. Auto disruption will also have a place at higher levels of play.
One thing we noticed about this game was that most forms of defense were possible in some way. The best way to avoid defense is quick cycles that maneuver around defense when traversing a cycle, and then scoring as soon as stopped. This minimizes the effects of the above.
One place that we believe may be overlooked early but could be an effective place to play defense is in front of the opponent’s source. While it is a protected zone, sight lines are significantly better for your alliance than your opponents. Every robot playing offense must go back to this location. It could be a very effective spot to slow down cycles.

[IndySam]

@sgeckler

Pneumatics - NAH! Love it!

Us 2 for only the second time in my 19 years on the team.

[Lexeo]

Yeah. We realized in 2022 that it was difficult to package and added a lot of failure points (especially if we tried to package it tightly).

With 20 PDH ports now, brushless motors not weighing much, and current limits trivial to set, it seems like most mechanisms that were formerly good candidates for pneumatics are just as easy to run with motors. There are no glaring things about this game that make pneumatics necessary, so we think we can just work around it.

[sgeckler]

In 2022 I made a conscious effort to challenge the team to NOT use pneumatics. It was so successful, we did it again in 2023. Going forward, I don’t see us using them again. It could happen, but I’ll do my best to avoid them. For this year’s game since we are going with an under bumper intake, there’s no need to deploy and retract an intake quickly, which is still the hard part of not having pneumatics IMO.

[suPURDUEperAndy]

This post is GREAT! And hilarious!!!

So many FRC drivers think defense is only one thing, smash, and there is really so much more. Watching good defense is so exciting (less so when it’s against your robot). When we are called upon to play defense, our driver usually walks away saying how much fun it was. Scouting for a defense bot is tricky, especially when some of the best drivers are playing offense most of quals.

[Lexeo]

Prototype/Concept Update

Over the past couple weeks we’ve been trying out a few concepts. So far we are still learning from the prototypes, and don’t have a clear path on what the robot will look like.

Intake

Like a bunch of teams, we were inspired by Team 95 expanding their frame beyond the swerve modules to have a full length intake. We decided to try this out on our 2023 robot.

Overall, we are pretty thrilled with the performance of this geometry. We have some work to do with centering and making the concept robust, but it feels very, “touch it-own it” in its current state. The layout of this can be found here.

“Cranberry” Shooter

Similar to the shooter used by Ri3D Cranberry Alarm, we decided to test shooting the notes from the sides, with significant compression (~5"). For this prototype, we are using orange Andymark 4" stealth wheels, with the flywheel installed. We are getting reasonably good results with this prototype, but we start to have consistency issues when we are far away. We made the compression adjustable on this prototype, and it seemed like ~5" worked pretty well.

We did test compliant wheels too, but we were spinning them beyond their rating. They expanded and started rubbing on the walls of the shooter. Even when running slowly, they didn’t seem to perform any better than hard wheels. We likely will not be using compliant wheels for shooters anymore this season.

“Quokka” Shooter

We wanted to try orienting the shooter wheels in a way more similar to the RI3D team Unqualified Quokkas. There is (probably) no way to get spin in this configuration, but some teams seem to be successful with it. Based on initial testing with the Cranberry shooter, spin doesn’t seem to help as much as initially expected.

“CAD in 12 Hours” Concept

We started laying out a concept similar to FRC 4522, Team SCREAM. This is also similar to the work done here.

There is a ton of flexibility here as far as the number of positions available. If things were extended with enough elevator stages, and climbers were positioned appropriately, we believe scoring in the trap would be possible.

“Lifting Hopper” Concept

We wanted to experiment with the concept of “removing” the note outside of the hopper to score into the amp and potentially trap. The nice part of this concept is the weight lifted on the elevator is pretty small. Likely 1 motor, and a roller or 2. It would likely require a 3 stage elevator to get the height we need for the trap (and still go under the stage) - so there is some complexity there. Both concepts are drawn with an “under the bumper” intake, but could likely be adapted to an “over the bumper” intake with some more thinking.

[JamesCH95]

Shooter looks solid, how does it do with worn notes?

The intake looks like a great implementation!

[Lexeo]

@JamesCH95

Shooter looks solid, how does it do with worn notes?

We aren’t far enough on testing that…but we’ll get there. Right now the biggest question is to determine the shooter style to put resources in. We started some consistency testing, but haven’t made any conclusions yet. We’ll hopefully have something later this week.

One of the first experiments we are doing is answering the question “does spin matter” using the “Cranberry” design. If the answer to that is no, We “think” the other style will fare better with used notes. I do have a few unused notes sitting there ready to compare worn to new once we feel like we are getting close enough.

[tjm1710]

/s

[Lexeo]

Updated Shooter Testing

We had quite a bit happen here and ultimately landed on a shooter style to move forward with.

Does Spin Matter?

The first question we had was “Does spin matter?”. The main draw to the “Cranberry” style of shooter was the ability to adjust wheel speed and get some spin in the air. We decided to shoot 12’ away from a whiteboard and mark where each note hit - using our “Cranberry” prototype (see earlier posts for more details). We varied the wheel speeds to what we felt maximized the spin from that design.

We ultimately concluded that there was not a significant advantage to spinning the note with the “Cranberry” Shooter concept.

Which Shooter is More Precise?

Using a similar setup to the above experiment, we wanted to compare the “Quokka” style shooter to the “Cranberry” shooter to get some idea of which is more precise. Based on what we saw here, the “Quokka” style is more precise.

The Verdict

Based on the results of both experiments, we concluded that we will be pursuing the “Quokka” style shooter. Our next few days will involve optimizing this shooter and determining exactly what it will look like.

Robot Layout

We met as a group to weigh out pros and cons to the “CAD in 12 Hours” concept and the “Lifting Hopper” concept. Both designs are fairly similar and equally capable, but we felt the CAD in 12 hours concept would be easier to pull off, since there are fewer hand-offs and we felt it would package a little smaller.

CAD Update

Now that we have the layout and the shooter style decided, our CAD team was able to get moving. We still have plenty of work ahead of us, but here is some of the current progress.

The goal is to progress far enough in the CAD by the end of the day next Saturday to be able to build the proto-bot. We will see how it goes…

[ehochstein]

Nice numbers on your bumpers! Looking good.

[Lexeo]

Battle of the Team Fruit

(written by Jersey, Strategy/Programming/Fruit-Connoisseur student)

After last year’s pineapple surprises in teams’ pits, teams 4272 and 461 have embraced the culture.

These pineapple surprises are a single team or multiple teams putting a pineapple into other teams’ pits for good luck! No one knows who does this pineappling, but we have our suspicions… Westside Boiler Invasion were pineapple-d last year at one of their competitions by this mysterious pineappler. Later they won that same comp with the fruity fury of the pineapple. We embraced the pineapple power and made it our team fruit last year; however, this year is different. We’ve decided to change our team fruit for the better. Sweat and tears went into creating our final decision vote.

After thorough searches on the internet and finding lore behind different fruits, we found the two perfect competitors: Ashley the Cantaloupe and Delita the Dragonfruit. They engaged in a Mortal Kombat style brawl (team vote). On one side, we have Delita the Dragonfruit, on the other, Ashley the Cantaloupe.

After a few pros and cons of the durability, cuteness, and the taste of the fruits, we came down to a decision. The great and mighty Delita the Dragonfruit came out on top with no one voting for the cantaloupe. Our dragonfruit will accompany the other team’s mascots this year! Be on the lookout for our new, fruitful mascots this year!

[UnofficialForth]

It’s cold… So cold.

We’ve used the last of our kerosene to light one last lantern in hopes someone might see us, come to us, save us from this forsaken place. Save us from… it.
We started this journey with high hopes, blissfully unaware of the blasphemy of combining bionics and beasts. Driven by the desire to create a new robot that drove completely by itself, we started down the path of the shark tests. Our computer fell asleep before the robot stopped driving. But we couldn’t sleep. Sleep as it circled us was certain doom. We survived only from snacks of the Dorito variety.

This wasn’t enough for us though, our greed drove us to fuse vision and odometry, creating a monster capable of always knowing where not only when it was stopped, but at every point. It could track us even when we weren’t in sight, it stalked tirelessly. It just wouldn’t…stop…looking at us…

So… cold…