Hello and good day to you from episode 50 of our podcast series Project Breakaway. A metaphorical and literal time in the day when we here at Predator Cycling take some time away from working in the back shop to come and share with our listeners what we're doing, how we're doing it, what it takes to do it. Our ideas, our innovative success stories and even our missteps and failures. If you find yourself with an interest in bicycles, composite manufacturing, out of the box design or even curiosities beyond. I encourage you to stick with us. Settle in and learn a little. I'm Courtney B, co-owner and project manager at Predator Cycling. I'm here with my partner Aram Goganian, the other co-owner, CEO, lead designer and engineer and boogie. Dancer is the boogie is the dance. I guess. Anyway, we went to downtown. To Broadway in Nashville. We are just east of Nashville. So going downtown to Broadway is like an adventure for us. I mean. Yeah. So many people and we are so old. Okay, maybe not that old. It's fine. It's fun. But when I look around and I'm like, oh, these people are all like 18, 19, 20 or bachelor at parties. So that could be any age. Yeah, I guess. I don't know. They were like, oh, we're playing classic music, rock music. And then it's like Blink 182. And I'm like, oh God, world. Okay, yeah. That dates us. But yeah, I know it was fun though. It was a good time. Yeah. Pretty nice. Anyway. Yeah. We had a good time. I mean, I don't know if I was a boogieing, but. No, not last night. It takes a special amount of alcohol to boogie these days. Anyway. Um, so. The last podcast, we talked about making stock component parts. And bikes and we were just mulling around the idea. Um. Yeah. You've been working on a mold for a stock major pilot track bar. We have. And today you pulled the mold for the hand grips. Yep, we did. Pull those out today. So for these, it's not like bar tape. It's like an actual molded like silicone type material. Yeah, it's it's like a rubberized material. Um, that are actually you've actually simulated in addition to like the bar. So like, why? Because I'm obsessive. I mean, what is a normal track bar? Is they do bar tape on track bars or they actually do like those grips? Those like rubber grips. So there's there's like three options you can really do. You either have, they do make like bar grips, like, you know, mountain bike styleish grips. Um, they also just sometimes run no grips and they just run um, um, like grip tape. Just the bar. Yeah, it's like just like they put like a strip of grip tape to have some resistance. Well, they wear gloves, right? They do wear gloves. Most of the time. time. Um, and then the other one is just bar tape, but they usually for the track they use very, very thin bar tape. Is what they do. Why? Oh. It just. I thought I had an answer in my head. And then I was like, no. Makes no sense. Don't worry, it's a conversation in my head about aerodynamics. I know nothing about. Go on. Okay. So anyways, um, yeah, that's usually how they play it out. So, um, and, you know, me being me, us being us, uh, we simulate. No. You being you. Okay. Me being me. Um, we, I simulate everything. So I actually modeled out some bar tape a while ago to actually see what the what happened to to the flow. Bar tape or a bar grip, a handlebar grip. A bar tape. On a bar. On one of our bars. If you bar taped it. Because most people do. Because the way we shaped our bars, you can't actually put a grip on. You have to put bar tape. Even bar tape would be like kind of goofy overlaps. It would be bad. Um, so like it's not. Anyways. It there's not a lot of really good options. The intent was always to make a customized grip. Now that we're looking at making a stock bar, it changes the perspective a little bit. So anyways. Are there other bars with customized grips incorporated? No. Are you sure? Relatively. I mean, I got not testing it. I don't have an answer. I just I just don't know if it's easily like Google or So there's no, okay, so there are, there have been custom grips in the past made for like aero bar base bar units for like levers and things. But they've always been kind of like molded to the hood type grip system. Like not removable. Some are removable. Ours are ours is essentially a grip, but it's designed specifically for the bar. Um, and it's Right, you couldn't slide it off and put it on another bar and like have it work. Right. It would not work. And also ours are like ours are very special because ours are left, right, specific. Yeah. And they're also specific to um, they're not molded like a lot are where they're molded like a tube. Ours are actually molded in the shape of the bar. And you've simulated all this. A lot. So what's the benefit? So the benefit is is that one, it allows us to create a more aerodynamical position around the actual property. Like we you can only do so much in carbon fiber. You can't make any single shape you want and then still have the integrity of the carbon fiber intact. So the idea was was to optimize for carbon fiber for aerodynamics for the structural integrity of the part, but then also if we could make a an enhancement to it because with with um, with, you know, different types of rubbers and stuff, we can actually make more precise geometry. We can mold the geometry more precisely. We could make some really interesting gains in aerodynamics from the grip itself and also um, make it more comfortable. So what are the gains you've seen in simulation? Because these uh, uh, grips are eventually going to go to like real life wind tunnel testing, right? Yeah, so there What are the obviously that hasn't happened yet. But what are you expecting via simulation testing? So we should see a pretty well, it's it's Okay, so if you're doing a static test at like between 30 and 35 miles per hour. Like you're not moving? Meaning it's a direct headwind, zero degree headwind or zero to two degree headwind. So straight on headwind. Which on the track is relatively what you're going to be seeing. And you're on track. Oh, I guess indoors. Indoors, you're going in a circle, it's relatively stable head-on wind. Because you're going into basically, well, a slight tailwind. I mean, they put an AC vent like right above. So they do, they usually rotate AC in the circle of the track, so they run AC in the direction of the track. To help you. Because as the cyclists start running, like racing, you will create your own wind, like your wind. You will create a movement of how the air travels through the building. So I'm sorry, is the AC vents going against you or are they pushing you with? They should go with you. What's a headwind, a tailwind? Tailwind, it'd be a tailwind. But it's not, I mean, I wouldn't really call it a tailwind. Because it's like it's so. It's not that helpful. No, but it's not. What happens is instead of it getting like usually. You'll see the velodrome get faster throughout the day. Because as the races start happening, the air flow gets going in that direction. I mean, we're talking, you know, hundreds, thousands of a second. That's what you're talking about. But that directional flow gets going. So what a lot of velodromes will do is they'll actually run the air ducks in that ventilation, that direction. So that that happens more organically. Mhm. So anyways, my point being is is that you're in a track condition, you're going into a pretty much headwind. Is what you're kind of going into. Zero degree angle. Um, so because of that, um, we can get away with more um trickery of wind. I mean, like we can do more. Wind trickery. Wind trickery. Um, so. Okay, so you've simulated this. They're supposed to be super, super fast, like assisting you. What is the deal with the UCI regulations? So there's not very many regulations on grips themselves. Well, but I meant like. So they're fairly thicker. And they do not exceed the dimensional restrictions set by the UCI for bars. Right, because we had this discussion because I didn't know anything. I was like, there's like narrow, narrow. I guess you want to be as narrow as possible. So they had like, because we printed some bars and I was like, holy crap. What's the point of those? They're too narrow. I can't even hold on to them. Yeah. But then you then there was like restrictions on. Where is it? The the base of the. And how the base of the bar, how wide it can be at the base of the bar. You're saying that you're counting the hand grip as part of the bar. Yeah. Which we don't know if it's. Well, it's a question because the the ruling is it says for track, it says is um the base bar. Is the is the translation is base bar. So we're writing the line until they make a new rule about hand off our. Well, it's it's unclear. the rule. And I've spoken to a couple of people that are like from different federation and and. They're also unclear of how it's being implemented exactly. And racing has been apparently from what I've heard. not incredibly consistent, but they're measuring it. So. I'm base bar typically you would refer to as an arrow bar. base bar. Like that's usually what it means. Not a track bar. So it's kind of misleading like the way the UCI phrases it. They say base bar. It's like, well, what's a base bar? Is it is it regular bar base bar? Usually it's. for time trial bars, base bar. So we just decided that we were going to stick to the 35. centimeter rule. of the base would be at least 35 cm, 350 mm or 35 cm. Um. with. um including the grip. And ours is a molded grip, so we can be very accurate on that number. Well, it looks super cool, the one you just pulled out. We want to make the other side. and then throw them on some bars and see what. see what happens. Yeah, it's be cool, so usually we try pretty hard to get like secondary testing as well. done other than just our biased opinion on something. So these are going out to get some secondary testing done. So we'll see. Well. Well. Um, that's just kind of what we were doing yesterday and today. So let's move on to the actual like main topic that you wanted to discuss. So. Um, we've this is going to be I think the first of what I think will be a few podcasts about um our. National Science Foundation grant proposal. Yep. So, just like a quick overview. Um, we're always coming up with like creative ideas. And we are a fully 100% self-funded company. Yep. So deviating from our normal normal business income is a stretch sometimes when we have crazy ideas that may or may not work. Anyway, someone's like, hey. you should apply for a grant. Yep. So we looked into it. So basically. Um, we are in the realm for this grant of agile robotic manufacturing for small businesses. Yep. So the point of a US grant is to initiate funding for like an idea and our case. um manufacturing processes that will benefit the future market. Or increase the US economic competitiveness worldwide. Or improve public scientific literacy and contribute to stem in general. Mhm. So, now you basically had an idea. one of many, we chose one. Um, that's never been done or proven and so basically the process is. you apply with an elevator pitch, which we did a few months back. And then you receive a formal invitation for a proposal, which is where we're currently at. Yep. And then I write a bazillion page proposal explaining your R&D models, uh the broader impacts of the goal, the intellectual merit, along with budgets and previous works, etc, etc, etc. Yep. Like a million different things. Yep. In hopes of receiving a small grant. Yeah. Relatively small. Yeah. Large for us to get a a kickoff project. For sure, it'd be helpful, definitely be helpful. And that's a phase one. So then in phase two, if you get selected, you continue on um to finalize like a beta operation type of your project. Mhm. And um the end goal is a phase three funding, which would be basically to like market your project, gain independent funding, and then bring something awesome to the market. Yep. So that's an overview of like what we're talking about. Yeah, when we say we're applying for a grant. Okay, so now. Our project is not in the realm of bicycles. Well, but. Yeah. More so we're using our business as the guinea pig for a project to prove like manufacturing workflow processes that we've used. Yep, for our bicycle products. Yep, exactly. So we've developed these processes along the way for the past few years. Um, but we would benefit incredibly from some funding that we ourselves cannot self-fund. Right. Kind of we've gotten to the point where we can't go much further. So we're in development of creating a workflow process for our smaller 3D printed components. And you, being the simulation design genius that you are. He's setting the bar pretty low. You have devised an idea in which AI, artificial intelligence, uh digital twins, which you can explain in a minute. And the metaverse can all work together in harmony to create optimal output. And minimal cost all under the guise of robotic AI predictive models based on synthetic data collection. Now. That's a mouthful. Explain. Okay, well, to like. Okay, you got to I think I think to explain the concepts. So what we've kind of modeled this internally. We called it IO work. Is what we've called it. And I know we've mentioned it in the past. Um, but we've never really kind of described. Our product is IO work. That's the name of the product. That's our end goal phase three. We sell it to you. IO work.app. Yes. Which is also a website you can go on right now and check out some. Yeah, there's some content on there. IO, sorry. A Y O W O R K. And IO is Armenian for what? Uh, yes. Yes. So it's like, yes. Work.app. Yep, because it loves work and it loves data. So proceed. Okay, so on that backing. Um, so basically what happened is is as I mean, pretty if anyone that knows about us. Or listens to our podcast knows. Um, modeling data simulation is like my thing. love it all. Um, so what we started doing is we started designing years ago. All of our bike models and assemblies are fully parametrically modeled. So meaning, um, if you update information on the on a sketch or part of an assembly, it'll update the entire system. And it also, to our point, actually updates molds, it updates tool path. It updates everything. So we had this like interconnectiveness. And what we wanted to do is is basically, um, which we've talked about before, is taking that digital process as far as we can. So taking that digital process to cam data that goes directly to CNCs. And then pulling data back from those CNCs. Um, running simulations of composites and then having those results of real time real products return back into the system. Um, and we wanted that like data connectivity. So we basically designed something, simulate it. And then link that process into the manufacturing process. And then validate that manufacturing process back to the data. And then if we're doing all that, we can synthesize our entire manufacturing workflows. We could then synthesize the the business side of it as well. And overlay everything and basically build a functional digital twin of an entire company. Um, and what kind of like, we never really thought we could kind of do it. Until we started on the RF 20 project. Because we had to build our own press for that bike. Because the press that we needed didn't exist or was exponentially too expensive for someone else to make for us. We made it ourselves. And when we built it, we built it so that it's fully data connectivity. So it has two-directional data streams. So we basically meaning we can from within an app, push data to the machine. And we can also retrieve data from the machine. So if we need to update cure temperatures, update PID values, um, anything that we need to do like that. It's bidirectional. Um, which let us down this path of what kind of IO work is becoming. It's turned into. Um, and like Courtney said, our first kind of, um, beta project of it, that the first attempt of what we're going to try. We're trying to do with it. To fully automate it. Is to do it for a print farm. Because we have a handful of small parts that we print. Um, and the, um, the the ROI math on it is is relatively simple. The demand curves and stuff are relatively simple to kind of do. Um, and it's less steps than like a more complex thing like one of a composite part. That has multi-steps and lots of physics and stuff and simulation involved. Um, but basically the idea is we can take data from both business data. So like, um, sales records, accounting information, um, traffic on a website. Um, demand, what we call call demand velocity is what we're kind of like internally labeling it as. Um, and figuring out that to what inventory you need. So it's basically, um, imagine lean manufacturing. synthesize into an equation. Um and then run an AI model and and synthesize with synthetic data models. Um is basically what we're trying to kind of do. The step that we take it kind of one step farther than um we take it than that. is basically we actually integrate that by directionally into our print farm. Okay, so that's what I was going to say. So AI and digital twins and the Metaverse, those already exist in manufacturing. Yeah, uh yeah, kind of sort of. It's it's. So you see the videos of the robots in like Amazon and you see them digitally on the computer doing a protocol or program that someone is like, you know, input arm rise up, input arm grab thing, input arm come down. Or in like BMW and car manufacturing. Yes. So what is the extra? Okay, well, well, two things. One. Most of that right now is what is either used like, I mean, from BMW, they've at least publicly stated that most of that is done for visualization and understanding of how the facility would work. Um, I know that Amazon just released now that one of their, um, um, new robot systems is, I believe, was trained in the Metaverse. Which is basically using synthetic data modeling. Um, what we haven't seen yet. is that connection. Where we're using real data off the floor to then sample set to build synthetic data models. And then reference those models to the real models. And that connection. So there's not like a human telling the robot what to do, the robot's learning as it goes. And via the live stream data that's coming in, like. It's. If you use sales data or web flow traffic, so we get an influx of people going on our site and they're really interested in speed play wedges. Yeah. But we only have 10. Let's say we have 10 on our shelves. Yep. Well, they've hovered over and clicked on that button. 100 times. Yep. So the robot would automatically know based on the data that it's collected from the web sales traffic that it should just go ahead and start printing more speed play wedges on its own without me saying or telling it to. Yes. So it's making it's increasingly it's. It's it's rechecking its predictions constantly. So like, you know, anyone that's run a business knows. That there's a lot of things you got to make gut calls on. Like it's. There's not we we build reports, we build models, um, you know, depending on how much time we have and access to data we have, how complex those models are. And then we kind of sit in front of the data and say, okay. We're going to do this. And you're usually making that decision on old data. When I mean old data. Like either hours, days or weeks old, sometimes months old data. And you're comparing it to models that are usually not very incredibly accurate models that you've built. It's, you know, Excel spreadsheets. Some quick books records. Like you're not using very complex models to build it. We're basically saying that what we'd like to do is give you highly specific models with predictive modeling on it and then synthesizing the outcomes of it and giving you highly accurate reports. And we can take it even a step further and basically build rulings, like rule systems into it. Where we can let that system actually make manufacturing updates in real time. Okay, but a phase one in particular is we're just like creating the linkage hub system. Like that's the phase one. So yeah, there's basically. Yes, and what we're trying to do is. It's a research and development phase. Yeah. We're creating this beast. Yes, and the hardest part of it is understanding is how it understands data. Because you're talking about so much different types of data. And um you're talking about sales data, traffic data, um meta data. You're talking about print data. STL files. Um, you know, printer efficiencies, printer outcome reports, um, you know, investment reports. ROI reports. All of that kind of data, it's it's. It's not that hard to actually just store the data. But to allow the system to figure out what the data means. And what the trends are. And what, you know. One of the things we keep internally we're doing is like we call it. You know, velocity, like sales velocity, product velocity. Like trying to figure out. That kind of stuff. And and so that. The the program can actually correlate the data well. Is the trick. Um. And you think you can do this? Yes, so I mean we've played with it with small data sets. And we're. Start small. Think big. I'm sure that someone is. I probably it sounds good. Um. Uh, but yeah, the idea basically we've we've we've shown that we can do it on a small scale. With data sets. Um, we now want to try and do it with bigger data sets. And we want to actually do it live. Like that's the idea. Is if we can get it to work. Um. You know, one of the cool things that's too about it is that we've actually set it up so that it's got a. Um an edge device as well as a cloud device. It's kind of the infrastructure we're going on. So it's it's pretty cool. Like it's. We we have uh, you know, very low latency local device that actually stores and analyzes data streams. And then also we push that to the cloud. There's a lot of little parts to it. There's a ton. There's a ton of parts. But I just wanted to use this podcast as like a introduction. Yeah, to the idea. Yeah, and and like I said, the end goal is. If we could take an entire business and make a digital twin of it and then simulate a business. Build synthetic data models of an actual business. Like that's the concept. That's where that's where. The idea is to take it, um, and our work would be the the basically the data collection. organization push pull system behind it. That's kind of the idea. So, Excuse Charlie there. Yeah. Having a little moment. You okay, buddy? He needs to get some water. Um, anyway, okay. Well, I just wanted to do the intro for the project. Um, because if we I submit it in a month and if we get, you know, selected as a Yeah. grant candidate, then you will be hearing a lot more about it. Yes. Once we get some funding. It'd be fun to work on. So, um, also, we are planning to send out um an interest survey to friends and like-minded business partners. To gather information and um kind of use it for a supplemental um market research thing for the proposal. Yeah. Um, so if this made your head hurt, I apologize. My head has been permanently fogged for almost six months trying to compile and organize your thoughts. I'm sorry. Anyway. Um, if you're interested in learning more about I work. app. You can go to Iwork.app. There's a couple videos um, Yeah, there's some information. We'll be adding more as we continue to work on the project. Yeah. Hey, anything else? Uh, no, but if anyone also is interested that doesn't get an email for the survey, we'll probably put a link in it in the podcast description. Oh, yeah. So you can. Um, you can check it out. Um, it is appreciated if people fill out the survey. Yep. Everyone loves a survey. Right? Yeah. Who doesn't love a survey? Okay. Let's wrap it up. We thank you for choosing to take some time with us. And we look forward to future breakaways. Look for us on Instagram and LinkedIn, Facebook, Twitter, and in person here in Tennessee. We ask our listeners to please share, like, and subscribe. We're available on all major streaming platforms. Thanks for listening, have a good one and find some time to break away.
EpisodeApr 26, 2023 · 25:54
← Podcast
Project Breakaway with Predator Cycling
50: Let Us Introduce Our NSF Grant Project, AyoWork, EP. 050
Courtney and Aram of Predator Cycling introduce a key component of their NSF Grant Project, AyoWork: innovative, custom-molded handlebar grips for track bars. Designed for optimal aerodynamics and comfort, these unique, left/right specific grips have shown significant performance gains in simulations, adhering to UCI regulations for competitive cycling.
Loading player…