FRC 5414 Pearadox 2023 Build Thread

Hi everyone! FRC 5414 Pearadox, is proud to be joining the #OpenAlliance for the 2023 season.
First some background on our team. We’re from Pearland, Texas (about 30 minutes south of George R. Brown). We serve all 3 high schools in Pearland and have an incredibly diverse team. We started in the summer of 2014 and our first season was in 2015. Since then, our team has grown exponentially with close to 60-70 people on the team. In 2022 we had a good season coming back from covid, being finalists at both district events, winning Engineering inspiration at the FiT district champs, placing 9th in the district, and winning the judges award at the world championship.
In addition to our primary competition robot Pearadox has built at least one everybot starting in 2020, and alongside our primary robot we will be giving build updates on our everybot including common problems and fixes that we find to better help everybot teams that we don’t attend competitions with.
We ran a small summer task blog to trial run documenting things better and you can check that out here. You can find more information about our build space in that thread.

We will be competing at:

  • Belton- March 9-11 (Week 2)
  • Houston- March 23 - 25 (Week 4)

Thank you for taking the time to read our ramblings and good luck to everyone for 2023!

14 Likes

Energy Heros will see you at both events!

1 Like

Was wondering when you guys were gonna join

I’ve been ready to join for quite a while but we decided to wait until we had more team buy-in before officially joining. We’ve gotten enough support for it to where I feel like we can make something that can help teams and that is not just me doing all of the documentation.

2 Likes

Like Beter said, definitely the most intimidating part of it is producing useful documentation. We’ve been generally not shy about a lot of things in the past, and we’ve attempted to do various blogs in the past unsuccessfully. Our build space is generally open to guests and people can always come see how far behind we are in person :smile: hopefully this will get more commitment from team members than prior attempts at various blogs.

Certainly being OA should help us produce better materials to take to competition!

3 Likes

8576 is looking forward to meeting up with 5414 and 5892 at the Houston event to see what cool stuff you all built.

This past summer, we constructed two additional robots in addition to our competition robot.

The first one was not officially named and ran ThriftyBot swerve modules on a small, lightweight frame using subsystems (shooter, intake, ball tower) recycled from the team’s old practice bot.

We took this robot to TRI, where it performed about as well as we expected considering it was scrapped together with old robot parts for us to test swerve for the first time. Our main issues at that event came from the old subsystems we recycled, not the modules themselves.
At one point in we did have a drivetrain failure. The entire bottom half of a module snapped off, wheel and all, we somehow didn’t notice until we got back to the pit and Allen brought us our wheel, and we were able to replace the module with spare parts we had with us.

We learned from that to make sure each module’s wiring was neatly bundled and detachable, for ease of replacement.

After TRI, we still had two events on the horizon Remix, and NTX. After seeing the various subsystem-related failures at TRI, decided to overhaul the old practice bot subsystems and start fresh. Having limited experience with extensive polycarbonate use on a robot, we decided to make this new robot’s subsystems almost entirely out of polycarbonate.
image
image

We originally intended to reuse the Thriftybot modules and switch to MK4Is when they arrived, however, the construction of the robot took so long that we weren’t going to finish the base by the time the SDS order shipped. We ended up just going straight to SDS modules. Compactness+lightness was the name of the game for this robot. We integrated the shooter and ball tower into one mechanism, and the final frame dimensions were 26" x 25". Additionally, we mounted all the motors on the base, running belts to various places in the ball tower and shooter. We took it to Remix purely to use Texas Torque’s build space, scrambling to finish for NTX.

At NTX, we were the captains of the 8th alliance and overall performed very well, being the 3rd best shooting robot at the event according to our data. We experienced cracking, warping, and bending with the MAXTube, but it didn’t cause any issues within the duration of the competition. However, after NTX while practicing on our home field the bolts holding the belly pan on where busted the tube completely open. 0/10 will not be doing it again
The SDS modules performed well. Having the motors in the base caused some occasional problems due to belting, but it was all fixable and could have been prevented with better CAD and preplanning.

If you have any questions about specific things we tried with our offseason robots (if it be swerve or how we ran our belts). please ask We’d love to answer any questions.
CAD can be found here

7 Likes

Saturday
We had a very successful kick-off, even finishing everything we planned on Saturday and Sunday early. We started out Saturday before the kickoff stream by reviewing our team’s goals that we decided on in the fall.

  • At least 1 every fully functional (legal) Everybot by week 1 competition
  • At least 50 hours of drive practice before worlds
  • Scoring prototype robot by week 2
    • Doesn’t need to be legal
  • Rapid fire strategy development
  • People enjoyed the year and team
  • People want to join back next year
  • People felt valued and that they contributed

We couldn’t really watch the game animation on twitch because of a poor internet connection, but we were able to go through the game manual and come back to the animation on youtube. We started by going through this worksheet to help guide our conversation about the game. Once we finished that we took a break for lunch. We created a list of all possible robot actions including all of the borderline sane ones. A personal favorite of mine is a drone that scores game objects. You can find the complete list of robot actions we came up with in our kick-off slides

We also did a quick and dirty max score analysis for auto where we determined that you can score a max of 63 points and that docking and engaging is worth the same as two high game elements scored.

While doing our game analysis we found a lot of similarities to the 2011 game and some other full-field cycling games like 2017. We also took a look at a TBA blog post titled “stop overbuilding. Just stop” to get a better idea of how a simple robot might be beneficial for our team.

To wrap up Saturday we broke into small groups to discuss what different level teams would look like (IE: what does a DCMP alliance captain do to meet that goal, what will the everybot do.), and then we broke down robot archetypes for things like cube only, and defense only. A complete breakdown can be seen in our kick-off slides.

Sunday

Our main goal for Sunday was to determine what success looked like for our robot. What did we need to be able to do for our week two competition?

Using slack we were able to gather a quick census from the team about what we think our robot needs to do. You can see these in our kick-off slides, additionally, you can see the slide I wrote about why we should do cubes only for our first event. One of our robot goals is to have a scoring robot by week 2 of the build season. This robot doesn’t need to be legal it just has to score on its own. To accomplish this we need to scale back our plans for our final robot and we determined that going into our first event we had to be able to score at least 5 cubes in tele-op and engage with the charge station.

Once we knew our robot goals for week two we were able to set our prototyping priorities and start rudimentary programming.

We tested the use of mechanical alignment on the base of the grid to ensure that our cone- or cube-scoring mechanisms can accurately place the game piece

we have some students pursuing a “tilted elevator” as a scoring apparatus to get cubes from the ground to the goal, targeting mid goal but the high goal would be a nice to have

Our cone-manipulation subgroup drew up various intake styles to be able to handle the cones effectively, and we established that it is going to be difficult to pick up fallen cones and that we should largely ignore them, but very easy to pick up cones from the human player “shelf” especially considering that the human player can manually line up the cone.

Some final thoughts about strategy that came up during our points analysis discussion, that we don’t think have been discussed elsewhere

Assuming you can do both cubes and cones, and intelligently place them for links, scoring high is only a 2-game piece advantage until you’re scoring 6 or more game pieces. In other words, if you can get practice scoring with a mid-scoring robot where you can do an extra 2 game pieces than a robot that can do high - you are outsourcing them.

Similarly, scoring 6 game pieces low is actually worth MORE than scoring 5 game pieces mid because of the extra link! And scoring 9 game pieces low is only worth one less than 8 game pieces mid.

Ultimately we think this points to the fact that the team should be focusing a lot more on fast cycle times rather than top vs mid vs low.

Also, this really stresses the importance of getting the dock in auto - getting 2 robots engaged is worth more than 3 robots docked but not engaged. So if you don’t need all 3 robots for the RP, it may not be worth trying to coordinate to get all 3 robots balanced unless you really need those extra points for the win - otherwise you might end up with 2 fewer points.

7 Likes

Oh man, day 6 already? I hope everyone is having a good week of prototyping. Here’s a lightbulb pear to brighten up your day.

The Ghost of Meetings Past

On Tuesday we began our prototyping efforts in earnest, and we’ve already learned a lot. We’ve tried mechanical alignment, cone grabbing, corb intaking, and corb shooting (“corbs’’ are cube+orbs - name credit to @dydx ). We developed and tested an over-bumper intake for corbs. They worked well with compliant wheels, but we still need to test a way to vector the corbs inwards. Using the Thrifty Bot Mecanum wheels didn’t work, but the larger 4-inch Mecanums seemed to perform better, so we’ll have to do a bit more testing.

We also prototyped a corb shooter; the accuracy is questionable at best (except for in this video apparently), but the results are promising, and we should have more details coming in our Sunday post.

Tuesday also marked the day that our rookie coders used one of the 2022 EveryBots we built to start trying to make the robot automatically move when at an angle measured by an ADIS16470 IMU and balance itself at the charging station. We started off with a simple P loop but modified it a good bit with some exponentials. We did find that over the time of testing, we definitely needed the capability to reset the gyroscope so we added a reset button.

As for our reacher team, they are still theory-crafting the best solutions to how to extend and get to the L3, working on potential arm/elevator designs. As they worked through a risk-reward discussion, they came up with some reasonable goals for how many points our robot should score by our first competition.

Wednesday was our big mail day. We got our order for a lot more cones for the field, some REV Ion stuff, and our Thrifty Bot elevator.

Our reacher team worked on assembling the TTB elevator. They mostly finished the base stage, as well as started the second stage. The project will be done by Friday this week, so keep your eyes peeled for that.

Our rookie programmers also continued to develop Charge Station auto-balance. Since we were seeing erratic behavior while using linear power control, we tried to find a non-linear formula that would restrict excessive power for larger angles. We chose the formula 1/5 * ^⅓, where x is the degree offset. We also ensured we’d have a minimum power applied since the robot won’t even move without some minimum (around 20%). Even after using this formula, we still had a lot of over-correcting causing erratic behavior. So we studied PID and implemented power-based PID control. We had minimal time to tune, so we reverted back to our formula for the demo and video. However, we also know that our bridge right now isn’t representative of the real field. We believe the double hinge design and the weight of the real bridge means it will “want” to be balanced. Our current build is light enough that it becomes out of balance very easily

As our designs become more and more definite, we find ourselves moving away from a mechanical alignment system. Due to a variety of reasons, we don’t believe a system like this will work for us, primarily because we found that having a slot in our robot where the hybrid zone dividers could fit into would often bind up and get stuck, while also failing to center our bot the way we intended.

Thanks to our wonderful parent and alumni volunteers, we now have a full-sized charge station, as well as two-thirds of a grid. However, we have a few issues with the charge station; it is about 60 pounds lighter than expected, is a major pinching hazard, and also doesn’t sit level most of the time. We’ve had more injuries and close calls than the 2020 and the 2022 truss combined and we’ve only had it for 2 meetings. We don’t expect the wooden version that FIRST provides plans for to be representative of the actual field charge station.

The Ghost of Meetings Yet to Come

Going into Friday and the weekend, we’re going to begin looking at better defining our robot architecture, and we’ll also host our first design review. We’re almost ready to assemble all of our swerve drives for our practice and competition robot. We’re going to keep prototyping our designs for this year and begin CADing and CNC’ing more sophisticated prototype designs.

When people ask me how to make a schedule, I like to answer that, by the time you begin, “you are already behind schedule and over budget.” For once though, I don’t think that’s true (knock on wood); we’re doing a really good job working on parallel prototypes and designs, and I feel that if the team gives a focused effort next week, we can have a robot working by our week 2 deadline.

3 Likes

By any chance, did you notice if the rocking of the center drop caused any significant negatives with getting the charge station engaged?

1 Like

not that I have personally noticed but we have yet to get a swerve robot driven up there to compare, I think the biggest detractor would be the moment when only two wheels are in contact which reduces the traction. But as I mentioned our charge station is not representative of the real one and we need to lay some poly carb on top for more accurate traction.

1 Like

behold an update, sorry for the little delay I got busy this past week and couldn’t muster the effort to get this pushed out the door sooner. Have a good week y’all and as always feel free to ask questions.

intake prototypes

We have tested around 6 different intake prototypes so far and the ones currently looking viable are all intended to be mounted onto an arm, elevator or a corb over the bumper intake. One intake has a pivoting “jaw” to be able to intake cones and corbs with the same set of horizontal rollers. Once it takes a corb, it is also able to shoot it out the top of the intake using a flywheel.

The second prototype has 2 rollers at the bottom instead of a jaw. The geometry is such that the corbs can reach the top roller and middle roller, and the cones can go point-first between the middle and low rollers. So that it can pick up fallen cones pointing at the robot.

We also developed two ground intakes. One uses a hinged roller to accept a variety of cone positions and corbs. The other uses two very compliant rollers made of surgical tubing stretched between two circular plates

Arm vs elevator vs shooting

After the prototyping that we’ve been able to do we felt that we could start our layout sketches and planning out potential robot designs for our Alpha robot. Our goals for this robot were to construct a robot and base as fast as possible to get as much field time as early as possible and get a feel for the game to develop game strategies. We felt that our previous prototypes could be consolidated into 3 different robots: an arm, shooter, and elevator robot. For each robot, we separated the prototyping groups into different subsystems and started a layout sketch for each robot.

elevator

The elevator had a dual game piece intake that had horizontal rollers on a pivoting “jaw” that had variable compression to either do fallen cones or corbs. The intake would likely be mounted to a vertical TTB elevator that has already been assembled. This would then be mounted on top of a SDS base constructed with MAXtube.

arm

The arm was also a dual game piece intake that was similar to the aforementioned intake. It had 2 horizontal rollers on the bottom that could be used to intake fallen cones. It then utilized another horizontal roller and the bottom roller to intake corb game pieces. We would be able to adjust what rollers come into contact with the game piece by adjusting the angle of the intake. The intention is that this intake would be mounted on top of a 1 DOF arm. We found that after some cad integration, we found that we could switch these designs out with the elevator ones

Shooter

The shooter robot concept is a 2022-style robot. We did extensive prototyping for shooting the corbs and were able to shoot game pieces from the outer edge of the charging station. We paired a simple over-the-bumper roller intake with the shooter.

After further investigation we found it was easy enough to build a drop-in shooter pod we could add to any of our flat bots and once we get the alpha bot up and running.

Alpha bot

The robot that we decided to flesh out and eventually manufacture was the elevator robot. Our team’s goal was to have a working robot by week 2. So we are hoping to start machining the robot at our next meeting in order to potentially accomplish this goal. The elevator robot will allow us to start developing auton paths and start drive practice cycling both game pieces.

|271x483.26460481099656

Swerve flatbed

This past week we finished setting up a Swerve flat bot using The Thrifty Bot swerve modules. After installing AprilTags onto the grid and a limelight on the flat bot, we worked on calibrating photon vision in different resolutions and testing them. We noticed some discrepancies with the XYZ values we returned, but the following day, we printed a new checkboard calibration image and actually secured it onto a board, calibrating the camera again and making sure to follow the guidelines on the PV docs. This improved our data significantly. We managed to get the WPILib SwerveDrivePoseEstimator working and take in data from Photonvision and AprilTag using their PhotonPoseEstimator. Basically, now we can look at an AprilTag and it’ll correct the robot’s pose quickly depending on how off we originally were. We also discussed the scoring process for teleop and we think we want to use this pose estimation to be able to just automatically align ourselves in front of any “column” of a grid (preferably like the closest one anyway if the driver drives the robot roughly close to the grid they want) with a click of a button. We will do more testing in the following week with alignment using this data and the pose estimation.

Cone drop testing

Cone drop testing has already gotten a lot of important attention on Chefi Delphi. We were able to get a human player station built up and did some cone drop testing. Something interesting that we found is that we were able to get the cone to land vertically almost 90% of the time. We likely still have more testing on this to do but, one of our programmers found that if the cone is slowly lowered and held onto using the opening flap until the middle of the ramp, it will land upright almost 90% of the time. Our field is not representative of the real field so take this as you will.

elevator

We continued work on the two elevator prototypes that we started on Tuesday of last week. As well as building up the rest of the elevator. While making the carriage, we learned we had cut the 2x1 that made up the length of the carriage half an inch too long, which meant we could not put the spring or the string onto the elevator. On Monday, we attached the chain and the rope onto the elevator and placed Neos on the bottom of the elevator, later manually making the motors extend and retract the elevator. A group was able to attack some old prototype arms of our flat bot, but it’s in a state that still needs some work. The base group was able to get the base CAD touched up and expects a base by the end of Wednesday. Then during the middle of the meeting, we all came together a split into 3 different alpha bots, Alpha shooter, Alpha Arm, and Alpha Elevator. For Alpha Elevator we were able to work with the shooter prototype and test different ways to merge the Elevator and the Shooter, and also figure out what angle the mouth needed to open to intake both Cobs and cones. For Alpha shooter, they were able to prototype a way to add the shooter onto a base, similar to how our intake was in 2022. For Alpha Arm worked on a way to come grab items and place them on the poles using an arm. In the end, we presented our integrations, and each group was tasked with creating their Alpha bot on CAD.

7 Likes

Do you have any video of cube shooting? That sounds quite interesting!

You bet I do, these are about 2 weeks old and we’ve gotten it to shoot even further and more consistent. All these videos are from our smug mug and will try and see if I can find any more recent ones tomorrow when people are awake.
Prototypes - Pearadox 5414
Prototypes - Pearadox 5414
Prototypes - Pearadox 5414

3 Likes

I’m curious what caused the MAXTube to warp?

Primarily crushing the light tube. I am trying to instill the mindset that we should almost never through bolt tubes in general without something to support (even 3d printed), but our team breakdown is:

  • 20 upperclassmen
  • About 20 sophomores
  • 30+ rookies (primarily freshman, but some other aged rookies too)

So it is hard to make sure best practices are always followed for the moment while we’re trying to go quick.

Edit: I’m excited to use the max tube, but I would not use the light variation in the base. The regular or max pattern are totally appropriate (as @JamesCH95’s analysis proved in another thread)

This is spicy! Now I think the top meta will be tiny little short bots that just bloop game pieces where they go

2 Likes

Yeah we just assembled swerve with 2x1 (the 1/8" side wall and 0.04" top and bottom wall. I think its really neat and worth the premium.

1 Like

Bring back 2016 bottle flip trend but with cones

1 Like

lol, we did a little cone shooting but nothing conclusive yet. Maybe that’s something I should look into more, it would undoubtedly be interesting.