Team 254 Presents: 2023 BREAKDOWN - Technical Binder, Code, Q&A

Team 254 is proud to present our Technical Binder and Code for our 2023 Robot: BREAKDOWN

Please feel free to ask any questions about the team, strategy, or robot and its code. Note that we do not publish our CAD, but I’ll be happy to provide screenshots and explanations of subsystems people find most interesting.

115 Likes

Can you folks reconsider publishing your CAD for 2024 right after Champs, so that we can work on a 254 clone for the summer? :grinning: :call_me_hand:

39 Likes

Cool tech binder - a few questions:

  • Do you know how much better supporting the falcon’s shaft is on the MK4is? How much of a difference does it make?
  • Do you use 3D printed pulleys or aluminum pulleys, or a mix of both?
  • What did you use for the calculations of how fast the elevator and laterator go?
  • What hall effect sensors did you use and how did you wire them?
  • Your behind the bumpers video said you used bushings for your intake but I don’t see anything about that in your tech binder. Were they referring to the ones in your elevator? If not, how did using the bushings work out and how much weight savings are they?
  • Has your dead axle construction methods changed since 2022 and if so in what way?
  • What current limits were on your drive falcons?
  • If you could re-design your robot, would you have the forks?

Feel free to answer as little or as many of these questions, thank you!

7 Likes

I don’t think I need to comment on this thing’s beauty. :100:

I have read everything, but aside from the quality and performance, this bot feels a bit different from the previous Decepticons we saw. I thought 254 usually went with the most capable option regardless.

  • How did you decide on intaking and scoring both objects from the same side?
  • Did you approach this game a bit differently?
  • After the season, is there any design that you wished you would have tried rather than your original?
12 Likes

I noticed you guys used a 2D unscented Kalman filter for filtering x-y position corrections. For the state and measurement models you used (dx/dt = 0, y = I), there’s actually a linear Kalman filter with a closed-form solution (WPILib’s pose estimator class uses that for performance reasons).

The things you changed in your custom copies of KalmanTypeFilter.java and UnscentedKalmanFilter.java were oversights on WPILib’s part. I just opened PRs to fix them (allwpilib#5830, allwpilib#5832).

5 Likes
  • Do you know how much better supporting the falcon’s shaft is on the MK4is? How much of a difference does it make?
    – We did it in 2022 so did it again for 2023. Many teams did not do this either year and had no problems with Falcon V2, we probably would find this necessary if we were using Falcon V3s.
    – I predict this will not be needed when using Krakens and their better shaft design, which is our plan for next season

  • Do you use 3D printed pulleys or aluminum pulleys, or a mix of both?
    – This year we tried to use more aluminum pulleys because we had the teeth shear off in a printed 18T 5mm HTD pulley in our 2022 Serializer at Chezy Champs, and it was crazy hard to disassemble and swap.
    – When printing a pulley, we printed in Solid Fill and additionally added ~qty4 1/16" holes for shear pins to go between the layers

  • What did you use for the calculations of how fast the elevator and laterator go?
    – Elevator
    image
    – Laterator, load is set high to be conservative accounting for the cascade rigging effect
    image

  • What hall effect sensors did you use and how did you wire them?
    – Hall sensors are wired as a limit switch to the Falcons and serve to zero the joints
    – For the floor intake and Laterator they are on the retracted hardstop which is the starting pose
    – The elevator starts midway up with a little kickstand for faster initial auton placement, but its Hall is still at the bottom of travel so only is triggered in a floor cone pickup pose or if we lose our encoder position and want to rezero
    – Forks carriage triggers a hall when all the way climbed, to tell the motor to stop pulling

  • Your behind the bumpers video said you used bushings for your intake but I don’t see anything about that in your tech binder. Were they referring to the ones in your elevator? If not, how did using the bushings work out and how much weight savings are they?
    – They are generally referring to the 3D printed blocks (Markforged with thin teflon tape on the inside) that served as linear bushing blocks for the floor intake extension. We were inspired by 125’s robot revealed early and were very happy with this intake deploy method, maybe will be the new standard versus a 4-bar for us???
    – The elevator uses 3/4" OD, 1/4" ID radial ball bearings, like many teams have built elevators with in the past
    – The Laterator tried to save weight by using just 3/8" OD, 1/4" ID PEEK bushings, that roll on dowel pin which are pressed into metal endcaps. The PEEK bushings are much lighter than steel ball bearings and saved a lot of weight on the 3-stage laterator

  • Has your dead axle construction methods changed since 2022 and if so in what way?
    – Deadaxle for floor intake roller exact same as we did in 2022 and others have implemented in year
    – The roller claw wheels are also on deadaxles, 3D printed pulleys with wheels pressed/glued/bolted-on have embedded 3/8" ID bearings and spin on 3/8" round standoffs both-ends-tapped to 10-32, held between the plates

  • What current limits were on your drive falcons?
    – Deferring to a software person to answer

  • If you could re-design your robot, would you have the forks?
    – I think by the championship / Chezy Champs, we found less use for the forks as most drivers were good and could triple balance very fast.
    – We ended up not removing them for Chezy because they don’t hurt now that they exist and if anything help add some weight to the back of the robot to prevent tipping forward.

31 Likes

Wow, thanks for the fast, detailed, and informative response!

2 Likes

Who says the robot isn’t the most capable? I felt pretty good about our OPR and performance at Chezy for example. :smiley:

Jokes aside, I do think having the intake and scoring be on the same side of the robot did make the autonomous much harder than we initially expected. We are still getting good at swerve autons, and the combination of spinning while driving proved to be tricky. I think this is one thing where other robot architectures are better.

However, in teleop, we optimized our robot around slamming into the feeder station because we believed that the extra driving time was worth it if it meant a consistent, repeatable, simple cycle for the driver to do over and over. No chance of human player messing up a dropped cone, etc.

That feeder station slam decision led to the vertical elevator in the back so that we could have an “up and in” pose, most other teams had to sneak up on the station because of an extended arm or angled elevator. The vertical elevator then led to the 3-stage Laterator, which had to be slightly tilted upward to get the extra height for high cones. The Laterator then led to the all-in-one intake that could do both objects while being lightweight, and excelled at centering cones from the feeder station.

Overall I think we are pretty happy with this robot architecture, albeit it different from the standard amongst other top teams, I think it was highly performant.

37 Likes

What are the black boxes by each of the swerve modules?

What was the reasoning for running cascade on the laterator over continuous rigging?

So you mentioned in your strategy that you determined non-upright / sideways cones on the ground to be something you weren’t going to pursue. While I agree that was the right call, what discussions did you have (during kickoff or later weeks?) that led you to this conclusion? Was it as simple as trying the different cone orientations and realizing it wasn’t worth it, or a more fleshed-out process?

3 Likes

I think you are referring to our simplified model of the CAN boards that we have on the bellypan to serve as CAN interconnect points. We use these for each swerve module and also before the elevator, laterator, and roller claw; basically each connection point.

5 Likes

Can you go into detail as to how the forks were rigged with the dyneema? A drawing would really help. Also, how was the bot able to hold its position after it was disabled when a match ended?

2 Likes

I guess I’ll add a couple questions to the pile, mostly manipulator based.

  • How far did those alternate manipulators end up getting prototyped. Obviously there was what looks like pretty close to final cad, but were there close to final prototypes as well or mostly wood/still preliminary prototypes? Kind of going along that same line, how long before your first competition did you guys decide on the configuration you wanted to use in matches?

  • Another alternate manipulator question, what were some of the pros and cons of those other designs? I have a couple of my own guesses but it would be cool to have some insight into what drove the decision behind the final geometry.

  • In regards to your scoring sequence, did you guys run the rollers backwards to place cones or just push the cone down over the pole and retract the laterator, forcing the cone out of the manipulator?

Amazing robot this season, super excited to see what you guys come up with in the future.

1 Like

Is the elevator your team used a completely custom-made one or is it based of a COTS product?

Is is possible for you can take a picture of this?

What factors led to using just 1 motor on most subsystems when there were available slots for more?

1 Like

This isn’t a recent picture, but here is one where you can see them clearly, just not wired to the individual modules yet. (Image taken from Team 254: Powder Coating Oven For Sale)

10 Likes

Ah, okay. I was wondering cause he said

So I thought, they made a custom CAN Board lol

What kind of surface wear did you see on the carbon fiber on the laterator? How did the bearings play with the epoxy?