254 The Cheesy Poofs | Behind the Bumpers | Curie Division Winners

Added: 2026-05-12

[00:00:07]Everybody is Tyler here at the First Robotics Competition checking with the legendary 254. It's the [music] Cheesy Poofs incredible robot number one in the FRC top 25 and I think in many hearts as well, too, uh, throughout the first world. We're going to be giving you a full systems overview of this robot here. So much to dive into, uh, from their incredible software that goes into this, their awesome DPS with their shooter as so cool just to see in person finally for this and then of course through their intake and how they're indexing. We're talking about some of the challenges that they had going through the season as well, too. Every team [music] has to work on a permit so you're going to hear more about 254 and what they've gone through here on Behind the Bumpers.

[00:00:43]Ryan, there's so much to break down on this robot here, but let's start out starting from the bottom up with this hopper floor, the intake, all this stuff. Walk me through, uh, how all this has come together for Cheesy Poofs.

[00:00:52]Yeah, so this intake is by far the most iterated part of this robot. So we've had a couple of folks this year, mainly mainly, uh, robustness because we knew that after we started seeing all the like the countless bumper rebuilds, there's so many of them, that the one thing that would that we could differentiate ourselves from the others is robustness. If we didn't break, we could even if we were tied in scoring performance, we'd be able to outlast the opponents. So if we start down here with the with the racks, I guess. So these racks at our first event, they were 3/8 polycarbonate with with titanium teeth so that the teeth wouldn't chip. So those consistently broke when we played against matches. So before our second event, we redesigned them completely and we just went with a full like completely over designed them so that they will never break. So now these intake racks are two layers of quarter inch SR PP and one layer of eighth inch titanium. So these have been great for us. They've never had issues with any kind of breaking or bending. In fact, I think these are the same ones that we've been running for the last month. So yeah. And then once we moving on to the roller path here. So, we have one main 3-in roller at the bottom, which which is I get it's we call it the main roller. It rips the balls off the ground and up up these kicker bars here. These kicker plates are titanium and because when we made them aluminum, all they did was consistently bend. So, these titanium kicker plates hold these kicker bars that are optimized for low compression without without dead spots.

[00:02:20]Then moving moving up the roller assembly slightly, we have between our between our first and second events, we or no, sorry, between our second and third events, we added a second roller here. This roller's purpose is to keep the balls keep the balls that are in the hopper from getting in the way of the balls that are currently getting ripped off the ground. So, in combination with these with these uh idlers here on the on the hopper floor, we can pull balls and they will sit way well above where the intake wants to pull the balls through.

[00:02:51]And these intake rollers are dual driven by Kraken X60s.

[00:02:56]Something that we've uh been talking to uh throughout our correspondence season, we've had a lot of uh upper echelon teams talk about like the SR PP material. Can you just talk to us a little bit more about how it's been successful and maybe teams who haven't used it before, why they should consider that? Yeah, so unlike polycarbonate, SR PP has fibers interwoven. And what that means is or generally for polycarbonate around bolt holes, it likes to stress likes to concentrate especially when you're bending around those bolt holes, which is the main issue that we saw when we had the polycarbonate break. So, by having these SR PP, the fibers take a lot of that load around the bolt holes and it doesn't actually just crack at the bolt holes like polycarbonate does.

[00:03:33]It's also It's also lighter and more impact resistant than polycarbonate. So, it'll absorb more energy when you take when you take impacts on it and then it also weighs about 30% less. One thing I want to ask you too is that we've talked about you know, the robustness right of the intake roller, but your acquisition has been so good as well, too. What have been some of the keys to success in just acquiring the fuel in the first place?

[00:03:52]>> I think, yeah. Uh one of the one of the biggest ideas you'll see with this robot is just iteration. So, we got the first robot out incredibly quickly and we've been very much iterating every everything ever since. So, this this kicker bar geometry we've gone through so many possible combinations of things. And like this second roller here that like we've gone through I think four or five gear ratio iterations, too. We've just been through like throwing everything at the wall to see what sticks. And this is this is what we landed on. So, we've Again, like we've done so much testing, so many like slow-mo videos, and we've put like GoPros inside the robot to see where balls get stuck. And we've done that to both optimize our robot for current draw and intake performance. And I love that you mentioned that as well, too, because these are things that teams when they're looking at testing that they should be considering also in order to get great data as well, too. So, I appreciate that.

[00:04:41]Before we get to our next segment, we'd like to thank the following.

[00:04:44]Build your alliance with so many other FIRST alumni who go to Kettering University. Every student at Kettering experiences their cutting-edge co-op programs that seamlessly blend the professional and academic worlds.

[00:04:54]Kettering co-ops are a fully immersive working experience at the leading edge of industry. Head on over to kettering.edu/first to learn more about their incredible programs and to get more information. [music] The next The next part in the ball path here is our hopper. This hopper is both designed for again robustness and ball storage. So, these are the So, on here on our bumpers we have we have what we call the bumper hopper. It's just It's a full 8-in polycarb that extends upward from our bumpers to both increase the width here and reduce the number of moving parts or not moving parts, just parts that we have on our robot here so that there's fewer interferences between things.

[00:05:27]Moving on to this front section here, we have a full carbon fiber hoop here that provides rigidity to the structure without without being super heavy.

[00:05:36]One other thing we noticed is that when these corners on on these front corners here, when these used to be polycarbonate, they would I guess they would just crack a lot. So, we we bought these uh carbon fiber corners here that we that have been served us great. They provide a ton of structure to these corners without actually costing us much.

[00:05:54]Something else I want to get to here is this net. This net has allowed us to go from about 50 balls in capacity to close somewhere close to 75 to 80. So, it's been a huge improvement to our robot this season.

[00:06:06]Moving Moving on a little bit more, once the balls are in the hopper, they get pulled in by the hopper floor here. So, the hopper floor we went with the through it through a series of prototypes, we decided on the the West Coast Products rubber back timing belts, which provide I guess more continuity in how the balls flow through versus having to like bobble along rollers. So, one very interesting thing about our hopper floor is that its pivot is linked to our intake deploy so that it's the the intake stows under the hopper floor and when we're trash compacting, we call it, when we pull in the intake while we shoot, the hopper floor lifts up so that the gravity can help us get the last couple balls out.

[00:06:45]So, [music] if we can push the intake in here, you can see you can see that the hopper floor lifts up and we can fully stow the intake under the hopper floor.

[00:06:56]All right, Zach, so we got the bumpers off now, so we get a little bit better view inside here. Let's talk about the feeder mech in the shooter and showcase how all this works. Yeah, so our feeder and shooter on this robot, this is the main thing that hasn't changed that much on it. So, the main things that have changed are mainly the geometry between our If we can pivot it up slightly. So, on this robot, we wanted to achieve both anywhere in our AZ being able to shoot from there and being able to shoot cross court as [music] efficiently as possible. So, if you notice here, the main thing that um the main thing that describes our compression variable our compression values and the positions in which we have everything placed is if I could have the ball, please. Yeah.

[00:07:39]Is the point in which we have zero gap in between everything. So, if you notice right here, this is what we call our triangle of compression, I guess. And so, what we do is we have contact with both the main drum, our bottom feeder roller, and our top hood roller. So, what this allows us is for us to have zero wobble as well as um a lot of spin-up time with our actual balls. So, the main reason that we have the consistent shot is because we are able to get all of our balls to the same uh speed and velocity on the uh throughout the entire shooter. Early on the season uh when your team's team started to go with the double roller up and then also have like an articulating hood, I think you know, a lot of people saw that as very ambitious [music] uh for what the season was. You know, we've seen what it proves to now, but early on we were looking at this, were there any difficulties in trying to package all this together? Yeah, for sure. So, the main thing with this architecture is we want to have as much effective width as possible. As mentioned in the intake and the hopper, we pretty much have zero things affecting all um our effective width. So, we [music] have things such as our igus on the outside, and we have all of our gear ratios and everything on the outside also. So, if you want to come to this side, um um for example, to uh because of our width, we only have an inch and a quarter on either side of our feeder plates. So, because of that, we had to opt out for things such as not being able to fit one pulley plus a gear. So, instead we opted out for having a double-sided belt, which is something unique to this robot so far. Um so, uh what we do is we have a wrap around here. So, this um we have the belt outside driving the main drum and the belt's interior driving the back hood rollers to get that reversing effect. Also, um throughout we had different iterations of our [music] backing feeder. So, um for example, we used to have it as poly, but now we have cat tongue if you can see. And the main reason we have that is because we did see slipping with our poly, so we decided to swap that with cat tongue on the entire [music] surface. And then if you want to come to this side, um as I mentioned before, um we tried to keep all of our width as effective as possible. So we have all of our shooter motors, which is not uncommon anymore, obviously. But we have all of our shooter motors nested behind, and that has [music] helped us also.

[00:09:58]As a team, uh when you saw, you know, your robot got, you know, released and then played in their first event, and then all of a sudden you start seeing some of these other robots take inspiration from your How does it feel, you know, being a member of the Cheesy Poofs and just seeing all these other teams take inspiration from your robot?

[00:10:12]Honestly, we were kind of surprised [music] cuz we didn't really think that this would be the leading architecture for being copied. We honestly thought it would be turrets. But I guess because we performed so well, we are blessed to uh have that.

[00:10:25]Before we get to our next segment, we'd like to thank the following.

[00:10:29]FRC Tees managed store program is an easy and profitable way to fundraise for your team or off-season event. FRC Tees will handle everything from production, packaging, bulk or drop shipping, payment collection, and accounting.

[00:10:40]Teams can set their own prices, and FRC Tees will take care of the rest. Fill out a quote request or get more information when you schedule [music] a callback at frctees.com.

[00:10:49]So you mentioned that you were a little surprised and that you thought turrets might be the way to boost have a plan for going turret at all?

[00:10:55]Initially, uh we were planning on going turret, but then we decided to look into the main things. So in uh the beginning of the season, we all have our kickoff build, which is a 12-hour long build in which we decide what we want to do leading on into the future. So we looked at the pros and cons of everything. Our main thought process on going bumper instead of a turret was because we looked at our um max ceiling and our max potential with both options. So with a turret, even if you have zero gap in between the balls as they fly through the air, we predicted that the max BPS you could get would be around 18 or so, which we have seen from some robots at this event.

[00:11:32]However, if you introduce multiple streams and multiple paths for balls to go through, that gets exponential over time. So, for example, we looked at a two width wide, we looked at a three width wide, and even a four width which was which is what we have right now. So, because of that, even if you introduce a gap between your balls as they fly through the air, we predicted that our ceiling for this architecture would be around 30 to 35. So, based off of those pros and cons, we determined that this would probably be the best plan of action.

[00:11:58]>> Where are you sitting at right now as a championship team? Um as of right now, these balls are fresh. We're probably around 25 to 28 is what we would say, yeah.

[00:12:09]A lot of the great software that goes into this robot here that really make it happen. Just pass it over to Max [music] to detail more about these things. I know we're going to run a systems check and check that out also, but let's start talking about what really has gone into this robot software side.

[00:12:20]>> Yeah, for sure. I'll start with the vision and specifically start with the cameras which is basically the basis of any good vision subsystem. [music] So, we have four cameras on this robot which is the most that we've sort of ever done. We have two cameras facing this way on the shooter side and those are actually 120 FPS cameras. All of them are grayscale. And then the ones on the side are 60 FPS just because they aren't used as much. They are helpful in some situations, especially when we're around the middle of our field and during our autonomous routine. I'll talk about it in a second. But yeah, these two cameras are doing most of the heavy lifting here. They've been really really great in helping us basically orchestrate everything that we do on this robot. Um from shooting to passing and basically our the core of the autonomous routine. [music] Um but basically the coolest thing that we've done this year is we've implemented our completely custom vision stack. [music] We're using a Jetson Orin Nano which is right there if you could see. And basically these four cameras are hooked up running on that one Nano. And we have actually a GPU accelerated AprilTag detection this year which has allowed us to support these 120 FPS cameras because there's a lot of processing that happens [music] and it's something that is not possible with traditional algorithms. Um but yeah, we have these four cameras running on the single Jetson and it's been absolutely great for us. We've had no performance issues. We're consistently getting 100 FPS plus on these two cameras on the shooter side. Um and yeah, in terms of what we're doing, um we're pretty conservative with what we accept. [music] Um but because we get so much great information from these cameras, it's really easy to get a good localization estimate. Um we were using multi-tag only actually because we have so many cameras and basically our only blind spot is on the intake side, which we're never really using anyway to um see anything. Um we are able to throw out any single tag estimate and only look at the multi-tags, which are both more accurate and more consistent, um which have been great this season.

[00:13:57]>> [music] >> Um another interesting thing is that we're sending packets over UDP. So, all the solving happens on the Jetson side.

[00:14:03]From then, we send um all the the poses and a variety of statistics over to the Rio, where we perform our [music] our filtering based on fields like re-projection error and ambiguity. Um and so, yeah, basically we send them through UDP just because it's faster. Um we don't have to worry about backlog or stuff like that. Um but yeah, it's been it's been great for us. [music] It's been the core of this robot, honestly, I think, and it's allowed us to get such accurate shooting.

[00:14:25]Um speaking of shooting, um I think our heading control >> [music] >> uh has been really interesting this year what we've sort of done with it. Um all that we're doing on this robot is based on like a predicted pose, right? So, we have a look ahead, and then we're also using feedforwards for heading based on where our driver is moving at the time that we want to shoot. And what that's allowed us to do is effectively align with um not only precision, but great speed as well. And depending on where we are on the field, we actually have an intelligent center of rotation. So, instead of always rotating based on the center of the robot, we can actually select a corner of the robot to pivot about in case that we're in front of the tower or next to a robot or currently getting bumped. So, that way we'll basically have the fastest alignment time towards the hub, and we'll be able to shoot as quick as we can, um because we're not shooting on the move uh due to some current limits and, you know, just it's not really practical for us this year. Um but definitely it's been helpful [music] in sort of being able to shoot as quickly as we can and avoiding any sort of defense. How was that implemented? Was that something for championships or did you have that at DC MP?

[00:15:22]>> Yeah, no. This we've worked on this literally the last week. So, the last week has been focused on faster alignment and then also the smart center rotation. I'll just because we know that the defense is going to be really intense here at champs. Um and we wanted to prioritize that. Especially as a number [music] one seed, right? You're really expecting a lot.

[00:15:34]>> Exactly. Yeah, we can't shoot on the move. So, being able to shoot quickly and um precisely is very important for us.

[00:15:39]Uh but yeah, I'll talk about our autonomous routines. So, this year we're using Corio like we did in 2024. Um we like Corio for shooting games just because it allows us to, you know, have these these paths, [music] these optimized paths um that we know that we're going to go for the say like the whole time, right? We're not really doing any on-the-fly computation like we were last year. Um but with these paths, we basically have three stages, right?

[00:15:58]We're doing the full two swipes and then that third swipe is sort of like a not a real swipe. We go into the middle of the field, try to get as many balls as we can in order to set up our driver um for success when teleop starts. Um but yeah, these these paths we basically have a conservative to greedy level on how much we want. Um and we like [music] to be as as greedy as we can initially just to beat that first swipe, beat the other robots on the other alliance, and get as many balls as we can. Um but the really interesting thing is based on whether we get knocked off. Like we have a system to see if we get knocked off our trajectory, um if our localization pose is too far away, we'll actually go onto a recovery [music] path. Um and you know, the whole point of this is to we don't want to empty a full hopper of balls if you know that we're not going to score. So, by able to, you know, if we go into the middle of the field and we get some balls, we get knocked off our path and we can't really shoot at that time, we go to a known recovery pose and then from then we go on a separate path back to collecting more balls [music] and then we can continue with our stage two and stage three. Uh just make sure that we are at least getting some balls in the hopper because we don't want to just be wasting balls.

[00:16:53]Um but yeah, I think the the vision in the autonomous routines really been critical in ensuring our success for this robot. It's been great for us this season. Absolutely agree with that as well, too. It's been awesome to see your team uh compete and succeed on the field. We can't wait to see, of course, as we're filming it's only a couple matches in, how this is going up as well. So, 254 Cheese Booth, thank you so much for taking the time. Can't wait to see more about this team coming up here, but stay tuned and can't wait to see how you do. Thanks a lot.

[00:17:16]>> you.

[00:17:18]Thanks for watching. Don't forget to like, subscribe, and click the bell to stay up-to-date on future fun videos.

[00:17:23]FRCT's managed store program is an easy and profitable way to fundraise for your team or off-season event. FRCT's will handle everything from production, packaging, bulk or drop shipping, payment collection, and accounting.

[00:17:35]Teams can set their own prices, and FRCT's will take care of the rest. Fill out a quote request or get more information when you schedule a callback at frcts.com.

[00:17:44]Build your alliance with so many other FIRST alumni who go to Kettering University. Every student at Kettering experiences their cutting-edge co-op programs that seamlessly blend the professional and academic worlds.

[00:17:53]Kettering co-ops are a fully immersive working experience at the leading edge of industry. Head on over to kettering.edu/first to learn more about their incredible programs and to get more information. [music]