11260 Up-A-Creek Robotics | Behind the Bot | FTC DECODE Robot

[00:00:02][music] Hello everyone. Reporting today for fun Robotics Network. I'm Abas and with me here today are Grayson, Owen, Arnov, and McKenna. And it is my distinct pleasure to interview team 11260, Upper Creek Robotics. They have been absolutely insane this Decode season. Jackson division winning alliance captain at the Houston World Championship control award winner. They're now here at MTI looking to do even more. And there is so much going on with this robot in the hardware, the software, every single subsystem. I can't wait to jump into it on Behind the Bot.

[00:00:35]All right, guys. So, first question has to be about the overall design approach.

[00:00:38]You know, you guys had the Clip robot last year and this year, this robot is just so fantastic in so many aspects.

[00:00:43]Walk me through how you guys set requirements for each of your subsystems and what you consider. Yeah. So, at the beginning of the season, right after the game releases, we want to brainstorm as many ideas as possible, crazy or not, and uh we imagine [music] what the game will be like at a higher level, and we want to be designing uh in order to allow ourselves to have the ability to modify our robot or to compete uh at that level. Yeah. And speaking a little bit, like there's so many different subsystems and they look pretty like modularly assembled, I would say, like distinct. Do you guys have one person designing everything? Everybody works on a different subsystem. How does that work?

[00:01:20]>> Yeah, so we use a modular design process to build our robot. This allows all of our hardware members to be working separately on different subsystems. Uh, and we use a full CAD model of our robot to make sure that they're able to be integrated.

[00:01:32]>> I see. And as far as like the Rev build system, you guys have been running that for a very, very long time. What advice do you have for teams looking to use the system? How can they get the most out of it?

[00:01:42]>> Yeah, we've been using the Rev system for quite a while and we've been finding great success with it. Uh recently we switched to using a go build a servo power module. Uh but we've been finding uh great success with RAB.

[00:01:54]>> Before we get to our next segment, we'd like to thank the following. True competitors know that every second counts. That's why Ketaring University challenges you to dive in right away as a first year student. Participating in robotics programs helps Ketarine students secure a valuable co-op.

[00:02:07]Whatever your interests, Ketarine gives you more space to work faster and win faster. Learn more at ketarine.edu/first.

[00:02:16]Let's let's jump into the intake right here. You guys have that full width intake. I want to if we can get the robot turned on uh intake like three artifacts for that like sideswipe cycle that you guys are so good at in the far zone. I want to first start with these rollers up here. Are they custom? Are they cast? What are they?

[00:02:30]>> Uh yeah, we got those rollers off Amazon. They're made of silicone. Uh and we found that they were really nice and grippy to hold on.

[00:02:36]>> And I see you guys have uh like these inserts like spaced throughout the rollers. What What are What is the function there? When [music] did you add them? Yeah, we added those uh near the beginning of the season. We wanted to have uh support uh inside the rollers.

[00:02:49]Got it. And talking a little bit about the screw I see over here. Walk me through what it does and how you made it.

[00:02:56]>> Yeah, so this screw is 3D printed out of TPU and it's uh it helps prevent jams with uh a couple other subsystems in our intake.

[00:03:04]>> Yeah. Talking uh you know I I also noticed that the intake as a whole is like sprung. So at what point did you guys add that? Was that from day one and why did you add it?

[00:03:13]>> Yeah, so the sprungness is was very quick in the season. I mean, a lot of our intake is asymmetry. You'll notice um especially with having such a wide intake, jamming was a big problem very early. Um so the being pulled down was one of the very first mechanis one of the very first ways that we decided would be a good way to make it asymmetrical and pull in artifacts quicker. And then you can see further like the cork screw and then other asym like the 3D print right here. All these are different. We got out of them at different times in the season just to make it better and better.

[00:03:44]>> Yeah. Uh as far as intake speeds go, has that changed throughout the season? What are you running now? And has it just worked the whole time?

[00:03:51]>> Yeah. So, initially we uh modeled this intake to be run off a servo. However, we found that that would uh that was too slow. And so, we later changed uh to this motor.

[00:04:01]>> And why why the servo? [music] Did you want to use your motors elsewhere or why do that servo based intake to start?

[00:04:06]>> Initially, we thought we were going to be using a motor for our lift. Uh however, later on uh we realized that we wouldn't be able to do that. Let's let's jump into some software with this intake. One thing that I've noticed is very very effective is your guys' auto ejection after that far zone like uh pickup of the artifacts. How are you doing that? What sensors are you using?

[00:04:24]>> Mhm. Of course. So, we have sensors all around our intake system. Um a brain uh beam break up here and then u color distance sensors where we use the distance um part of that. So we have this data flow that tells us um how many ball numbers we have and we also look back at the history to avoid tripping.

[00:04:40]Um but with that we are able to in um [music] in addition to the path innings we are able to time out and also time out based on how many balls we have. So for example when we're intaking a line of spike in auton we're able to end it with the path but also end early if you already have um three spikes. So talking specifically about uh like any current draw sensors or after after you've collected let's say like four or five artifacts you just you run into there you have three in here a couple in here [music] what decides to outtake what what sensors are you using to make that decision >> of course so we have this whole algorithm for intaking but you can see how um the first the first bar and um this bar are separate so this bar allows us to keep in our three motor uh three balls but once we have our intake is full we'll run um will always ejects out so we avoid um overpossessing.

[00:05:31]>> So it's two separate motors for those two intakes. Yes.

[00:05:33]>> Okay. Got it.

[00:05:36]>> Before we get to our next segment, we'd like to thank the following. Studa Robotics is inspiring teams to build better robots with their new array of FTC team options. Check out their updated bevelgearss and Maverick hex shaft motors, planter gearbox options, and 6mm hex components and shafts for extreme power transmission. Go to sudica.com/root to learn more and apply for discounts.

[00:05:59]>> Let's let's talk about the transfer now.

[00:06:01]Moving up a little bit. Uh first, how are you powering it? You're already using two motors for the intakes. Is one of those also used in the transfer?

[00:06:07]>> Yeah. So, this motor up here is used for our intake and we have a motor back here with a chain uh that drives our transfer.

[00:06:13]>> Got it. Has that have you um you know had to deal with optimizing like the transfer efficiency throughout the season or has the speed been pretty good for you guys? Currently our launching speed is limited by our transfer but we are uh pretty happy with it. Okay. And uh what exactly is it like you're running the transfer at full speed? It's just like maybe the path the artifacts are taking or does it also have to do something with the software?

[00:06:34]>> Uh we are uh running it's the distance of the artifacts.

[00:06:37]>> Okay. I see. As far as compression goes in the transfer, this is something teams have dealt with a lot throughout the season. Uh how did you guys manage that compression and make sure that it led to as much shot consistency as possible?

[00:06:47]Yeah, we wanted our compression in our transfer uh to be adjustable. So, inside of our transfer uh we use uh these pieces of foam along the back surface uh and that um allows the artifacts to have enough compliancy they need as they move.

[00:07:02]>> Is that something you've had to like replace throughout the season or it's just been like set it and forget it?

[00:07:06]>> Uh yeah, we've been replacing the that uh foam consistently throughout the season. And uh we use also use TPU rollers as well.

[00:07:12]>> I see. The TPU rollers. Okay. They're kind of on the inside. Yes. Might be a little hard to see from the from the front. I can show the >> Yeah. Ne next biggest thing is moving up to the turret. You guys have had like a very very consistent turret all season.

[00:07:26]Uh how have you made sure that you're always always locked in? There's no like vibrations or moving off or anything like that.

[00:07:32]>> Yeah. So, we have these racks on the side on both sides. Um with having two on one each side, it really helps us stay stable. And then you can see that we have a lot of support for each gear.

[00:07:43]And then on our servo, this is all supported. We just added a lot of support [music] to every little mechanism and make sure that everything is nice and snug.

[00:07:51]>> Yeah. Talking kind of about your hood and the uh you know the base turret as well. Backlash is a concern we've seen from teams all season. Is that something you guys have faced? If so, how'd you overcome it?

[00:08:00]>> Yeah, we were facing a little bit of backlash. I mean, we have a lot of weight on the top of our flag. You can see. Um, and this was hurting us a little bit at the beginning of the season, but really as long as we were keeping everything nice and tight, it was not that big a problem.

[00:08:13]>> Yeah. To add on to that slightly, we switched to using a servo powered turret uh later on in the season to free up a motor for our launching. And with those servos, we realized uh that we we changed their offset in the servo programmer in order to make them fight themselves just slightly. And we found that that removed a significant amount of the backlash in the system.

[00:08:33]>> Yeah. Could you give teams like ballpark like was it like 0.01 even less than like how much are you changing these values just to make [music] sure they fight but not draw too much current.

[00:08:41]>> Um I don't remember off the top of my head but it's it's not a lot. It's a very small amount >> point4. Okay. Okay. Sure. Sure. Sure.

[00:08:48]That's that's fine. I also want to talk about the rails over here. These look like you guys have like custom uh rails for your rack. Uh was there a lot of development involved? Do you recommend teams do this or talk to me about them.

[00:08:59]>> Yeah, so the idea behind this is that we wanted a stable hood. Um, so because we knew that we'd want to launch from the far zone and have a consistent launching. So instead of making our hood be the adjustable thing, we have these three-stage extendable dove guides that they were pretty hard to design, but um I'll be honest, they work really well.

[00:09:15]We were very happy with them. They you can see on the inside they are um you can see their profile to the artifact ball shape.

[00:09:23]>> Okay.

[00:09:23]>> So we're very happy with them. They give us a very consistent path up into our flywheel and help for accuracy a lot.

[00:09:29]>> Awesome. Yeah. Now, let's talk about sensors for the shooter. Uh, I see we have the this uh we have one sensor right here. I see the lights coming from it. What is that sensor? How are you using it?

[00:09:39]>> Mhm. This is a color distance sensor where it's um a color sensor, but we use this it's the distance feature. Um we just use this to see how many balls um we have. So, we have a top, middle, and low. And that just tells us that we have a ball in middle.

[00:09:51]>> So, do you have three color distance sensors or only one that you're using to determine how many artifacts? So we have one here and then one on um okay on the bottom and um but for the top we use as our beam brake to determine but we don't use the color feature.

[00:10:07]>> I see. I see. Yeah. Let's let's talk about now the sensors in the shooter itself. Uh before talking about the encoder I see you guys have three uh sensors right here. What sensors are those and how are you using them?

[00:10:17]>> Yeah. So all three are beam brake receivers. We have one emitter over on the other side. And the reason why we have three is because obviously the artifacts have holes in them and we were struggling a lot with the beam going right through the holes. And even after adding a second receiver, we were still seeing that sometimes the beam can make it through both. So after adding a third one, we were able to angle the emitter a little bit. It was pretty hard just because of how like how straight the beams are, but we were able to figure out a way to get all three receiving the beam until an artifact was going through. And are you guys wiring all of those as separate sensors or are you uh you know combining the wiring for those [music] as far as the software is concerned? How are you dealing with that >> in software? The top is just um registered as one sensor. So I believe we're wiring wiring them together.

[00:11:05]>> Okay. Okay. Got it. Uh now talking about the encoder here, you know, we we've we've seen teams talk about using a separate encoder directly mounted on the flywheel for noise concerns or veloc like accurate velocity estimation, anything like that. Why did you guys decide to do this? Um it's just because the encoder that um the motor came with uh was not high resolution enough. So this um gave us a better reading out because uh the noise really mattered to how we were reading the speed.

[00:11:32]>> Are you guys having to do any like noise filtration or anything like that or how are you dealing with that?

[00:11:36]>> Yeah, definitely. We do a low pass filter for our speed readings.

[00:11:39]>> Okay. And did you guys just determine that threshold just like through testing? Was there any like really strong justification for it or you just recommend teams test? Um, we printed out the graph, so we just recommend teams to test that.

[00:11:48]>> Okay. Yeah. Yeah, that that makes a ton of sense. I I want to talk a little bit about the auto shooting algorithms you guys have here. Your guys' far shots are just absolutely lasered in every single cycle. What are the biggest factors in maintaining that consistency?

[00:12:01]>> Um, thank you. Of course. So, um, we read our position with our otos and also a camera here. We just read, um, the a weighted average of that. Um, and then because the OTOS drifts off throughout a match and the April tag reading is able to pull it um back with that coordinate, we actually draw a like virtual point onto the field so we know which way we're aiming because we're actually not aiming at the April tag but at the corner of the target.

[00:12:27]>> Um, with that we build a launch vector which tells us um which we um Oh, if you want to see this which tells us our distance from the an angle from the goal. with that um we can change our angle and uh flywheel speed. We used to we actually used to use a formula for the speed and angle um we just that we took from different points on the field but now we just do a lookup table. So it's a lot easier to change and um recalibrate for m uh different >> as far as getting the data points for that lookup table. Is this like 10 points 20 points? How many shots are you taking in order to generate this lookup table? Um right now I think we took eight points on the field there. Um >> three of them are in the close launch zone and then four others are um in the outer zone. We actually only we actually started with six but then we had to build a few more further ones just for the uh corner shots and pulling up the graph.

[00:13:20]>> For sure. As far as the shot speed itself, are you just launching the artifacts through as [music] fast as you can or do you have to be at the correct RPM every time or are you adjusting the hood? How are you dealing with this like speed and consistency balance?

[00:13:32]>> Mhm. Of course. So, as I said previously, we control our whole intake system and with the indexer is how we control um how fast we shoot. I know there's two windows for um our shooting.

[00:13:43]One is that our um turret or our a we are actually seeing the April tag. So, we know we are actually aiming at the goal and the second thing is um our flywheel speed. So, we have a error window that we're actually we're only able to shoot within. So, it segments the shots.

[00:13:58]>> I see. Let's talk about hardware with the shooter. I think one of the uh you know biggest strengths of your guys' team this season [music] is that top spin shooter. At what point in the season did you realize that's what you need?

[00:14:08]>> Yeah, of course. So, very early on, right after the game reveal, we started with a backspin, but we we knew that we wanted to be a farbot. As soon as the game revealed, we were like, "Wow, we need to shoot from that for this point."

[00:14:17]So, we saw that with the backspin early prototypes, the artifacts were bouncing out a [music] lot. And so, we decided to mess around with different uh spins. So right after backing we tried to top spin and we found that it was much more consistent with staying within the goal and then later we actually took off a lot of spin with our reverse roll or counter rollers and because we found that putting too much spin will hurt as much because you can't put if you're putting in power into spin you're losing power going into the actual [music] shot. So the spin helps with or the counter rollers help with that.

[00:14:45]>> So at this point now is it like a zero spin shooter slight top spin? What are we running?

[00:14:50]>> We still have some top spin because obviously you don't want a no spin. It's going to it can duckle ball and then it can become inconsistent. So, we're still running with a little bit of top spin.

[00:14:56]>> Do you know like approximately like, oh, your surface speed of the top is like 10% faster than your bottom rollers or anything like that or it's not not as important as just like testing it?

[00:15:05]>> We did a lot of testing to refine the speed. Um, that was also supported by a lot of math. I believe our uh top it does run at about 10% faster.

[00:15:15]>> Okay, cool. Uh, now another thing I want to talk about is this crazy autonomous program you guys have. Uh, just go ahead and explain it. I don't want to spoil the show.

[00:15:23]>> Yeah, sure. So, we have two main autonomouses. Obviously, we have the corner autonomous that can get 24 balls uh up to 24 balls. And this user this uses our camera, but how we're different from other teams is instead of making like a contour range and going there, we're able to make like a whole grid and we're able to find which a part of the grid has the most like pixels of the color we want. And that basically allows us to instead of having zones, we're able to go to that the point we exactly want to to give the most balls possible.

[00:15:48]Currently, sorry. Currently, >> and and then I know you guys also have another autonomous that's more like full field. what's going on there.

[00:15:54]>> So, that's actually one of our most unique autonomouses. Uh instead of having to use a ideal like sorter to sort the balls since we wanted to get the rank uh the ranking RP, something we do is we're able to shoot balls off the goal and it returns back into our intake. So, our team is aware that of the obelisk and the spike. So, thus we're able to know uh exactly the order of the balls in our robot. So, we were able to like bounce and basically change the order inside of our robot so we can sort and get up to nine balls essentially.

[00:16:20]>> Fantastic. Last couple things we want to talk about is the hang. Uh I you know we we'll I think we're going to see it run a lot more at MTI uh because of that increased threshold. Let's just see how it how it lifts first and then we'll we'll talk through some of the hardware there.

[00:16:35]Okay. So it's just a it's just a park right here. Walk me through how much space it takes up in the zone. Any auto alignment? What do you guys have going on?

[00:16:42]>> Yeah, so we have no auto alignment. You can see on the sides of the robot there's little taped marks right here.

[00:16:47]Um this is helps our driver know where we're allowed to be within the zone. And then we take up just under 9 in of [music] the zone. And you can also see we have LEDs on the bottom that illuminate the so the refs can see easier.

[00:16:58]>> I see. Uh and then the last thing is you guys played a lot of matches at Worlds, right? You went all the way to Da Vinci.

[00:17:04]You played so so many matches on Da Vinci. How did your driving strategy change specifically in that zone and what are you taking from there to MTI?

[00:17:12]>> Yeah, so as the driver, I saw a lot more defense than I was expecting. uh it in world level and probably here I'm expecting to see that we need to while playing [music] offense play more defense also. So, as a Farbot, it's important that we are putting pressure on the close bot when they're doing their gate cycles, getting penalties on them in that secret tunnel area. And then also, when going out to do our launches, when the other Farbots launching, it's important that we are putting pressure on Dawn by hitting them um when they're about to launch and making them missoupled. And what's really good about our team is that we're very consistent. So, as long as we're not getting hit, we're going to be scoring more.

[00:17:48]>> Yeah. And and you guys have a much larger and heavier bot than other teams.

[00:17:51]How did you use that to your advantage?

[00:17:53]>> Yeah. So, like I was talking about with hitting the other Farbot, we're able to [music] speed up and then it's really hard for them to stop us. Honestly, we're just able to ram through all that inertia.

[00:18:02]>> Yeah, exactly. And then you can you can see that in a lot of our da Vinci matches is on one of our strong suits is that we [music] were lowering the scores of other far bots by around 10 five to 10 artifacts.

[00:18:11]>> Yeah. Well, Uppa Creek, thank you guys so much. I mean, you just have consistently some of the most [music] fantastic sensors usage, clever system designs. It's just always a pleasure to interview you guys every single season.

[00:18:21]Share with the world what you guys are doing. Reporting for Fun Robotics Network, I'm Abh Hassaw. This team 11260 Upper Creek Robotics. Thank you.

[00:18:30]>> Thanks for watching. Don't forget to like, subscribe, and click the bell to stay uptodate on future fun videos.

[00:18:36]Studa Robotics is inspiring teams to build better robots with their new array of FTC team options. Check out their updated bevelgearss and Maverick hex shaft motors, planter gearbox options, and 6mm hex components and shafts for extreme power transmission. Go to sudica.com/roobots to learn more and apply for discounts.

[00:18:55]True competitors know that every second counts. That's why Ketaring University challenges you to dive in right away as a firstear student. Participating in robotics programs helps Ketarine students secure a valuable co-op.

[00:19:05]Whatever your interests, Ketarine gives you more space to work faster and win faster. Learn more at ketarine.edu/first. edu/first.