How software-defined manufacturing fits into real factory operation

0:08What's up, everybody? Welcome in episode 258 of the Robot Report podcast. I'm Steve Crow, executive editor of the Robot Report. Thank you, folks, as always, for tuning into the show. Joined, as always, by Mike Oitzman. He's the senior editor of our robotics group. What's going on, brother? Hey, Steve. Welcome back. We missed you, but we had a good time last week. We had Brianna and Gene, the three of us on the show. It was a great conversation. You guys don't need me. Welcome back. No, we still love to have you here. It's like they drove up and I flew the coop, right? Yeah. Thanks for holding the fort down.

0:47Yeah, I had a fun vacation with the family. But back in the swing of things, let's see if I remember how to do this. Hopefully it's like riding a bike, right? But throwing right back into the fire. Did you see I want to talk about this off the top? because this is one of the more interesting, I don't know if ironic is the right word, but stories that I've seen in quite some time. So Amazon drone delivery, we've talked about it before on the show. They announced earlier this week that they're expanding to 500 U.S. cities, their drone delivery services. And then almost just a few hours later, if anybody hasn't seen it, I've posted it on my LinkedIn.

1:23We're writing an article about it right now, but you can find it on all the socials, this viral video of an Amazon delivery drone dropping a package into somebody's pool. I think it's in Texas. Was it in Texas? And it's just, I don't know, man. It's I think it's kind of apropos for all of this talk about AI and robotics and how, how intelligent all these new world models and foundation models and how it's going to revolutionize robotics. I just think back, Amazon, they have unlimited money. They've been working on delivery drones since it was first announced in 2013 that they were going to do this.

2:03So they were working on it before that, but you can say they've been working on this for at least 13 years, 13 years. And, you know, it's not only that they dropped it into the pool and they, I think there was an X, a technical explanation for this. Amazon talked about it last year, actually in a blog post. And they said, their aerial imaging techniques often view pools as a flat surface, right? They can't see the depth down into the pool, right? They can't see it. So I think that's probably what happened here. But just, it's just also like how high they're dropping the packages from. It's that video and there's other videos that people have posted where it's their first time getting a drone delivery from Amazon, then there doesn't seem to be a whole lot of care taken into account for how they should drop these packages, right?

2:57And it's just wild. It's just wild that for something that's been in development for so long that some of these silly mistakes are still happening. So, but what were your thoughts when you saw this? It was one of the first things I think I shared with you guys this morning. Yeah, I think it is shocking. But at the same time, it just reminds everyone about, you The corner cases with AI are so – that's what we really get paid to deliver, right, is against the corner cases in addition to the mainline application. Right. This is exactly that, right? It put it where it shouldn't be. The system did all of the things that it could to determine.

3:41This was a good spot except for the pool finder feature. I don't know. One minor detail. Yeah. So some people were making jokes. Maybe they were dropping off a pool toy. Maybe it was chemicals to take care of the pool. It was a soft landing. Maybe it was a fragile object. And so dropping in the pool is a way to get your glass delivered. Yeah, right. And another thought that I had was that, because I believe the way that it works is if only certain places are eligible for this drone delivery service, right? And I believe you get, as the user, you get to select sort of the ideal location for where the drone should drop off at the package.

4:24And it's like, did this woman do this on purpose, right? That's true. To try to go viral, right? I don't know. But even if that's the case, it still shows a limitation for the drone technology, right? It should get to this location and see, hey, this is not the right place to drop this off. It's not going to work. And then, you know, maybe it has to go back to the warehouse that it took off from or try to find another location in real time on its own. But regardless, I don't think that happened. I don't think it was a scam by the woman who posted it. But regardless of how this happened, I think it shows that there's still some, to your point, edge cases that are pretty comical at this point.

5:06But on the other side of this, Steve, there's also opportunity. I mean, the point here is that this is still a problem that still needs resolution if we're going to try to deliver things to people's homes with autonomy. Look, I interviewed a new company last week called NoTip. They're trying to deliver food delivery to suburban and rural communities. So I was very interested to hear about this company. They want to do a dual robot system that brings an autonomous vehicle on the street. Then the little robot jumps out, takes the food up to your patio, jumps back in the vehicle and goes back to the restaurant.

5:44So interesting idea, interesting opportunity against still lots of hard problems to solve in that solution. And I told them, you know, this is a great opportunity. I think there's still lots of corner cases to solve here as Amazon just started a company that has unlimited funds, right? Let alone a tiny little startup who's got an innovative idea, right? So, again, we're here to support all y 'all in the audience who want to approach these. But, yeah, gosh, lots of difficult things to solve, as we just heard about. You said that startup's name was NoTip? Is that the name? No tip is the name of the company.

6:26Yeah. Love the name. Love the name. A little aside, but the tipping culture has gotten out of hand in this country. Yeah. Well, that's what they're doing. The whole business case, a$30 meal costs you$30 to get it delivered on Uber now if you have to drive any distance. So that's just not economically viable. No. No, for anybody. Right? Yeah. Interesting. Good stuff there. All right. Great show today. Our guest later on in the program is Roby Lynn, founder and CEO of R2 Labs. We've had Roby at some of our shows over the years. Was a contestant last year at the Roby Pitchfire competition. I believe they're based down in Georgia, so I've seen him a couple times.

7:07Yeah, he's part of RoboGeorgia. I've seen them a couple times at the RoboGeorgia events. So it'd be good to just set this up for us. What are we going to hear from Roby later on in the show? Yeah, well, Rob is going to take us through the whole, you know, innovation model that they are chasing. They are attempting to build an orchestration layer for factory automation. I mean, we're talking about classic manufacturing, where there's PLCs and other legacy systems on the floor, and really deliver, you know, modern AI based orchestration controls on top of that. And I think, you know, they put a ton of work into their first product here.

7:46And so he's going to introduce us to the product and what, you know, their business model, what really the problem they're attacking in classic manufacturing. So, yeah, great interview. Definitely got to go take a look at them if you're doing anything in that space. Yeah, great stuff. So we'll get to that later on in the show. And R2 Labs, they're going to be one of the startups in the startup alley at Robo Business. So if you like what you hear from this conversation and you want to connect with Roby or you want to learn more about our two labs, you'll have a great chance to do that. At our Robo Business Show, Mike, which takes place October 20th and 21st out in Santa Clara, California.

8:24This is the 20th anniversary show of Robo Business for folks who may have been following along closely. You'll see we redesigned the website recently. Within the last couple of weeks, my big ugly mug is on there for some reason. Just full disclosure. Not my idea. Believe me, that was not my idea. But we have some great pictures looking back at some of the companies that have been at RoboBusiness over the years. So one of the first ones that I noticed right away was Helen Greiner, co-founder of iRobot. She's in one of the pictures right off the bat. So it's going to be a lot of fun. There's some interesting things that we have planned, but we'll have over 2000 people, over 100 exhibitors on the show floor.

9:02Again, most of the speakers are posted onto the website and the agenda at this point. We have a few slots that we're still dotting our I's, crossing our T's on. So hang tight on that. But thank you to everybody who submitted, looking to speak at the show. So again, if you haven't seen the agenda, please go to robobusiness.com. We're going to have over 70 people on the stages over the course of a couple of days. And again, to be clear, what we're trying to do here is we are trying to bring together folks who are deploying robots at scale, who have built businesses for robots that have scaled and learn from some of the best in the business that have done that.

9:37So Rodney Brooks is going to be there. Ken Goldberg is going to be there. Jonathan Hurst is going to be there talking about the future of robotics to kick off the show. We're doing an AI keynote panel with Andrew Davidson, the chief AI officer at Universal Robots. Michael Vogelsang, he's the head of AI Foundation Models at Collaborative Robotics. We're going to have somebody from Amazon is also going to be there as well. We're looking to maybe add somebody else to that panel, but ABB Robotics is going to be there. So really focusing on companies that have done this over and over and over again, deploying robots in the field at scale and talking about these new technologies, right?

10:18How are they looking at AI? How do they see AI helping them right now, right? Not 10 years from now, not at a 53 % success rate. How is AI helping them build the product or perform in a real world environment 99 % of the time. That's what we're trying to do, right? So we hope to see everybody there. Again, you can go to robobusiness.com to learn more. The early bird discount ends on August 31st at the end of the month. So if you want to save a couple hundred bucks, we suggest that you get your tickets now. But again, go check out all the speakers. I think you'll find it's an amazing group of people for the 20th anniversary show of Robo Business, which takes place October 20th and 21st out in Santa Clara, California.

11:01Again, hope to see you all there. And if you have any questions about the show, reach out to any one of us here at the Robot Report and we'll answer those questions for you.

11:09Roby Lynn: All right, on to the news of the week, Mike. And we have folks covered with all the latest news and information with our network of websites, the Robot Report and Automated Warehouse. I think the big story of the week, obviously Unitree Robotics, the IPO happened earlier this week, Mike, and they absolutely crushed it. I think they raised close to what was over$900 million. They closed, the stock was up over 460 % at the close of the first day on the, what was it, the star market? Is that what it's called in China? So just, we have a story about this on therobotreport.com that we'll link to, but just an incredible performance, right, for an IPO doesn't necessarily mean that it's going to stay this way, but what are your thoughts on the early days of the IPO here for Unitry?

11:59Well, yeah, kudos to everybody at Unitry for what they've done in the industry. Arguably, one of the companies that shipped the most humanoid robots up to this point, again, into the early market for research. A lot of companies have bought the Unitry robot to train their models, the early models, as they were developing their own humanoids. So they definitely have profited from that. And the point here with all this war chest, you know, what happens next for them? I think that's my big question. What are they going to do with the money to expand their manufacturing, you know, build a team, continue to innovate as, you know, humanoids grow, but good for them.

12:45And I think it, you know, sets a mark. So now everybody else in the market that is considering an IPO knows what it's going to take to get to this point. And, you know, we'll continue to watch how they grow. I think as of today, they're up about 350 % gain over the IPO price. So it was up and down a little bit. And as everyone knows, you know, while this looks great from the IPO perspective, you know, nobody inside the company yet can cash out on this, right? They're probably all locked down for six months to a year plus before they can actually sell their stock. So they're incented to, or their stock options to hang onto the stock.

13:28So we'll have to see what happens both with the stock, but also, you know, with the company here, because the competition continues to grow. Right. So we'll see. Yeah, I think it's what I'm excited to see is how Unitary continues to do. Again, now that it's a publicly traded company, it's going to have to reveal financial information. Yeah, we'll get to see the real numbers now. Yeah, and we have some real numbers, right? In 2023, it sold just five humanoids. And then in 2025, it sold over 5 ,200, right? So I think it's great to have a company that makes humanoids. There are others who make humanoids that are not public, but this is a big chunk of their business, right?

14:17So even if, you know, when Boston Dynamics or Hyundai is a public company, right? But Boston Dynamics is such a small portion of the Hyundai business. They don't go in detail for that, right? They don't go in detail, right? Ubitech is a publicly traded company, but they make all sorts of things, right? Unitrade, I believe, is 100 % in on robotics, right? So they make droids, they make quadrupeds, they have robot arms, they have LiDAR units, they have other motion control components. But I think it's all for, or certainly the large, large, large majority of their business is geared towards robotics development.

14:58So I think it'll be interesting to see how this continues to grow. And then again, when Agility, if its SPAC closes later this year, or maybe it's the beginning of next year, depending on how things go. So again, it's going to be interesting to see how a company like Agility does financially because they're the opposite ends of the spectrum. And I think we've talked about this before on this podcast where people are comparing apples and oranges, right? Unitree has never, they don't strive to build a robot that works in an industrial environment. They haven't yet. They haven't yet. That's not what their goal is.

15:31I think their goal has very obviously been to build a low cost, performant humanoid robot for the research community and others. Enthusiasts. Entertainment. Yeah. Adrian McNeil, CEO of Fox Club. I know he has one. He bought one for the company to mess around with and test things on. That's what this has been about for Unitree for a number of years now. So I know some people have said, you know, their robots aren't built for commercial deployments. I think they're missing the point. I think there might be some sour grapes in some of those comments. Right. I think Unitry should be celebrated for what they've done for humanoids.

16:13Right. To get them, bring them and actually produce a product at scale that people can buy for a very for a low dollar figure for compared to other humanoids. And let's see. Maybe it comes out that they do want to build an industrialized humanoid. I don't know. We'll have to see what they do. But the speed at which they're able to iterate and introduce new products is kind of unmatched, right? I don't know how you can say otherwise. But I'm excited for that. I'm excited to finally see how they're doing from a financial standpoint. I think in our story, we reported that Unitree made a profit of about$41 million in 2025, up from$14 million in 2024.

16:56Can compare it to Ubitech. Ubitech lost about$104 million. Right. So here's a company that's profitable. What does it matter if they're building an industrial humanoid or not? They're not. And they're making$41 million. That's pretty good. Right. So I do think it's a missed opportunity for a U.S. company. And I think I said this around the time of the FCC restrictions, whenever that was a month ago or so. Why doesn't an Agility or a Fauna or a Boston Dynamics, clearly there's a market for a low cost humanoid platform that is mainly geared towards robotics developers. I had long thought that it wasn't viable from a business case.

17:38Maybe Unitary is showing that it is viable from a business case. I don't know enough financial details. Again, a lot of these Chinese companies are subsidized by the Chinese government. So I don't know how that necessarily plays into the puzzle here. I'm sure it plays a significant role would be my guess. But I think it also shows that maybe there is a market for a U.S.-based humanoid company to do both. Why can't they have a low cost, lower cost, stripped down research version? and then still have the industrial version. So I think they should be celebrated, but we'll see. We'll continue to follow that and see how the stock does.

18:15Another interesting story about a company going public, Mike, is Fort Robotics. They're going to take their safety stack public. And again, Eugene Dimitri covered this for us at therobotreport.com. If you haven't seen this yet, the interesting part, there's a few interesting parts of this, But Fort Robotics is going public via a SPAC that I think values the company at just over$550 million. So what was your take on this? Yeah, well, again, Fort, longtime sponsor, supporter of the Robot Report. So thanks for that. And the key here is that, you know, they have been growing. You know, in the article, we talk about revenue grows 62 % year over year in 2025.

18:58and operating expense. They've been able to keep operating expenses down, so it seems like they've got a decent set of books here. And they've got over 600 companies. They're all marquee names in the robotics industry across the board. Agility, Google, Cobot, Zoox, Carnegie Robotics, all those types of companies across the entire spectrum. So the safety products that they make are critical. to the successful deployments, which is the segment of the market that they've carved out. And they have rock solid products that perform, which is important. So I think they've got everything they need here.

19:40And likely, I couldn't get out of the article here in the research before exactly what they're going to do with the cash, But for them to be able to ramp up their manufacturing and to grow their reach and grow their channel, that's likely what they're going to do with this investment to continue rolling in. And again, it's good to have another public company that we can see the actual numbers for as well as potentially participating. So grab it and put it in your portfolio if you believe that the market's growing. Why not? You said the 600 companies, a couple that we've listed, right? Agility, Google DeepMind, Cobot, Zoox, River, Ocado, DoorDash.

20:27These are great companies, right? Companies that are doing things, hopefully, as far as I know, the right way, right? Carnegie Robotics, another one. So those are some great companies to have at your disposal. It says that the SPAC is going to bring in over$200 million in gross proceeds. So we'll see. Again, we've talked a lot about SPACs on this podcast, especially recently with the agility SPAC that we just mentioned and is supposed to close. I think either late this year or early 2027. These things always make me wary based off previous SPAC performances. I think it's safe to say this for you as well.

21:07I mean, we wish these folks nothing but the best. We hope it works out for them. But yeah, I think it's a little weary to see a company going public via a SPAC. Does that mean that the investment landscape's not as healthy as maybe some think it is? Is this not where investors want to put their money? Are there other reasons to more quickly scale things? But again, I'm a little weary of the SPAC thing, but we'll continue to follow this one. But again, nothing but the best. Good luck to everybody at Fort Robotics. Final story before we get to the interview with Roby Lynn. Mike, Logic is automating rail to road, transloading with the new Logic palette and digital twin operating system.

21:47Yeah, this is a story that we reported on Automated Warehouse. And I like this story primarily because of the kinematic, the actual robot. It's a powered palette. But the design is to roll these smart literal AMRs into the rail car, off the rail car, sort of the idea like the classic Kiva robot, although they're not picking them up. You're actually loading a pallet of things, a container of things onto these sort of three foot by three foot size vehicles that can then crowd themselves and organize themselves into the rail car. And then when it comes to unload, they come up to a platform. The whole things just roll off of the rail car very quickly.

22:35So it's designed to speed the unloading of the rail car, then get it onto a platform so that those containers, those small containers can then find their way onto another tractor trailer for the final mile type of delivery. So I just love the innovation that they have here in terms of solving that problem of quickly turning around a rail car, you know, to get stuff on and off so that they can keep the rail rolling, so to speak. That's a basic idea. Yeah, that's interesting. It reminds me, maybe not that solution necessarily, but I'm watching the video on the automated warehouse story of their logic palette.

23:19and that very much reminds me of a Slip Robotics product. Yeah, yeah. It's like a smaller Slip Robotic idea, a little bit more modular, the Slip Robotic concept. So Slip, you know, proved that that's an opportunity. And so here's someone who's innovated, you know. You know, another one it brings to mind, it's sort of like the Mitra solution. So Mitra is grabbing pallets and moving it into a big array. So it could be some potential opportunity for interaction in those types of areas. But yeah, just the most innovative stories that we talked about this last week on Automated Warehouse. Yeah, very cool.

23:59So we'll link to that story, all the rest. And again, we have you folks covered with all the latest news and information about the Global Robotics Network with our websites, The Robot Report and Automated Warehouse. Make sure you follow us on all the socials. make sure you sign up for our weekly newsletters so you don't miss a beat. Now on to the featured interview. Joining us today on the show, Mike, is Roby Lin. Again, he's the founder and CEO of R2 Labs. Roby and R2 Labs, they're going to be at the startup pavilion at Robo Business. We're going to have 10 startups there. We're giving these booths away to them for free, right?

24:33This is something that we've long done to try to support the robotics startup ecosystem. But what are we going to hear from this interview with Roby? Yeah, so I think this is a great interview. If you haven't heard of them before, we're going to learn about the capabilities of the R2 Lab solutions. It's basically bringing modern software-defined techniques to industrial automation. They produce what they call the R2 Autonomy Controller. It's designed to sit in a manufacturing control cabinet, act as a central platform to connect all the PLCs, robots, vision systems to the MES and ERP software and other IoT devices on the factory floor.

25:13And again, to connect it all with modern software today. So they're bridging the gap between traditional OT systems and modern IT software tools. So if that's a problem space that you work in, you definitely want to listen. And everyone knows that PLCs aren't going anywhere, right? They are fixed in time. They do the job. They do it repeatedly. They do it reliably. But, you know, there's a lot of room to modernize and centralize the orchestration of all those PLCs and other automation equipment on the factory floor. So let's give a listen now to this interview with Roby Lynn.

25:53Well, I want to welcome Roby Lynn, founder and CEO of R2 Labs to the show today. Roby, thanks for being here.

26:03Roby Lynn: Yeah, thanks, Mike. Thanks for having me. We've been looking forward to this for a while now, I guess, since Robo Business, right? Yeah, and we're getting ready for the next business. So it's been a while. And one of the reasons that you brought that up, of course, is that you participated in the pitch fire last year. That's where we first met you. I learned about the R2 Labs solution. And while you guys have been super busy the last year, I know, as you've been growing and building the business. So why don't we start there? Tell folks a bit about the genesis of R2 Labs and about yourself, of course, a little bit in that story as well and how you came to realize there was a need for what you guys are building.

26:47Roby Lynn: Sure, sure. Yeah. R2 Labs was founded by me and two other colleagues from prior experiences together. We worked in some combination of automation environments, manufacturing environments, tech startups combined. Right. And specifically, Sulise, who's our CTO, and I worked in a couple of different manufacturing plants before doing controls engineering work and automation engineering work. And we were, you know, we everybody came from Georgia Tech. So we're used to the, I'll say, software defined robotics world. OK. And then we jump into the manufacturing automation world, which is a different world.

27:32Roby Lynn: in a way, right? You know, things are done very differently. There's a lot less openness in terms of software, I'll say, you know, a lot more proprietary tooling, a lot more use of, you know, off the shelf things and like integrating stuff together and less writing custom software. So that was a bit of a shock, I guess, for us, you know, and so So we thought about that, you know, we kind of got, for lack of a better word, somewhat frustrated with limitations that we felt were placed on us as engineers with kind of the existing tools. And so we wanted to bring, you know, I'll say software-defined robotics type techniques to automation at a very high level description, right?

28:20Roby Lynn: So we're used to things like ROS. We're used to Linux. We're used to open source software. We're used to software engineering for robotics. And we saw that that's not really something that has a huge foothold in the automation side. And so we set out to originally see if we can make those techniques work and then ultimately develop a product around that that would kind of bring the best of both worlds to automation. There's a reason that automation products, I'll say, PLCs, gateways, all this kind of stuff. There's a reason why those things exist and they work well and they do their job. So this is not some criticism of that.

29:04Roby Lynn: I think this is more like a, can we bring modernization ideas to work together with those existing platforms to create something even better? So that was kind of the genesis. And so Sully and I and Rebecca all got together a few years back. I guess it was in 2021 and kind of started talking about it. Can we build something to fill this gap? And is this something that people are going to be interested in? And so that was an interesting part of our story as well. I guess, let me back up though and say, I'll start with where we are now and then I'll go back to like how that evolution happened. So our product is called the RAC, the R2 Autonomy Controller.

29:51Roby Lynn: This is a intelligence platform used for manufacturing in general. So this is a software and hardware product that can be used to orchestrate production assets, collect data from those assets, control those assets, and basically bring more modernization to existing equipment from the software side to allow those things to work better together, to increase throughput, increase quality, that kind of thing. So instead of relying on PLCs and PLCs alone, we can have PLC-controlled equipment and add our system to allow those PLCs to do things like AI-based vision processing, for example, or interfacing with IT systems, more simply, or orchestrating robots between those PLC-controlled machines and fleets of robots and that type of stuff.

30:38Roby Lynn: So it is a bolt-on factory modernization solution.

30:45Roby Lynn: That's effectively the overview of what our product does. It goes right into the control cabinet. Yeah, go ahead. You sort of have built this term you're describing as software-defined manufacturing now, right? Yeah. Take that concept a little bit further for us now. So, yeah, I think software-defined manufacturing, software-defined automation, these terms have been kind of floating around and maybe mean different things to different people. there's interest in I'll say software defined automation from a how do we program PLCs with something that is maybe more resembles computer software normal computer software more than ladder logic right and I think another aspect to this is like how do we allow those PLCs to really integrate tightly with software systems as a whole.

31:43Roby Lynn: So you have manufacturing execution systems, ERP, you have IT infrastructure that's been deployed. And we want to have production assets work in a modern and efficient way that is going to require them to be fully integrated with those IT systems. And at the same time, we have robots coming on the market all the time that are software native, okay? They're using MQTT and HTTP and kind of web ideas like this. That's not something that a PLC has typically had to deal with. And so there's kind of a bridge issue there. So there's gateways to solve that problem. Of course, some PLCs provide MQTT and HTTP interfaces and all that kind of stuff.

32:25Roby Lynn: But I guess my point is that there's kind of two different worlds. There's the OT world, the IT world. They live separately, but they don't have to. And so that is what our product is designed to be the centralized controller for. So I'll give you an example. We have a customer right now that does sheet metal processing and they were looking for a way to allow their machine to be intelligent enough to adjust process parameters based on detected defects. Okay. While also tracking what's going on with the production and being able to display that and log it to their MES. So they make a whole range of different sheet metal parts, but it's a very manual process.

33:17Roby Lynn: There's a lot of operators that cut and assemble these parts and hang them on a conveyor, which then the parts go through heat treat wash and then they get powder coated. And they're having a lot of issues with both tracking the production and defects in the powder cutting process to the point where if some defect happens in this production line in the conveyor somewhere, the part has to go all the way through the entire process and come out the other end before they know anything is wrong. And so they wanted a way to effectively, you know, visually inspect these online and then inform the production line, which is controlled by Rockwell PLC, inform the production line, adjust painting parameters, change this parameter, et cetera, to eliminate these defects before we start making a bunch of scrap.

34:08Roby Lynn: Right. So this is really the idea of how do we make our machinery more intelligent? And our answer to that, obviously, is you do that with software. So that system, we deployed our system in their plan. It goes right into their control cabinet, integrates with their PLC, integrates with their MES, provides dashboarding capability. And this is an AI-based vision application, obviously. But this is something that wouldn't have been, you know, feasible, really. this tight integration between the IT world, the OT world, the AI, you know, these cameras, etc. That really wouldn't have been possible with another tool.

34:50Since you brought that up, I mean, you're living in this space, as you just described it, between IT and OT. What do you think's been the biggest friction point? Why is it taking until now for someone like yourself to bridge that gap?

35:08Roby Lynn: I think there is a battery of solutions available for solving pieces of this problem. You know, so there are ways to have a PLC talk to MES, of course. You know, I mean, that's existed for a long time. There's a way to have a camera integrated in a production line. The problem that we have seen, though, is that these are all point solutions that are brought in by different system integrators, maybe, or developed internally or something like that. And there's a lot of integration effort to kind of make all the different pieces work together. So you can have system integrator A, add cameras, system integrator B, connect this to MES, you know, and then, okay, well, how do we get system integrator A's solution to talk to system integrator B's solution?

35:52And there's a lot of, you know, work to be done there, right? So our vision is, can we have the centralized platform that allows all this integration to happen in one place through software, allows for all these different workflows that the customer may want to do, whether that's, you know, vision, robotics, et cetera, et cetera, keeps

36:12Roby Lynn: that all in one place, driven by modern code that any software engineer can write. And don't make it something proprietary. I mean, there's, you know, when software engineers write software, they don't pay for a license to use the development environment, right, which is very different a lot of times in the OT world. So that was another thing that I think was difficult for us to accept. Like, oh, so let me get this straight. Not only do we have to buy the hardware, we also have to pay a license for the programming environment in order to program it. That is mind blowing for somebody that's used to Ross, for example.

36:55Roby Lynn: You know, so this actually leads into what I was going to say before, which is that we're engineers, right? Engineers like to build products for engineers. So, you know, we think we think, oh, we're engineers. We would buy this. Therefore, it makes sense to everybody. Not true. OK. As an example, I'm an engineer. I like cars with manual transmissions. That doesn't mean that everybody likes cars with manual transmissions because not everybody thinks like engineers. Right. So so when we first got started, we were really thinking about this from like the engineer's point of view, like what's attractive to the engineer?

37:31Roby Lynn: The engineer wants full programmability. The engineer wants the capability to write in any language. You know, they want, you know, SSH access. They want all this kind of stuff that really makes sense for a software person. But when you come into manufacturing, the customers aren't really so concerned about that, right? They're used to dealing with integrators. They're used to maybe not necessarily writing all the OT code themselves, having somebody else do it, blah, blah, blah. They're just looking for a solution to their manufacturing problem, you know? And so something that I think we came to along the way was how do we take these things that are very attractive to engineers and make this something that makes sense for the actual customer?

38:12Roby Lynn: This was part of our, you know, I'll say like, you know, customer discovery journey and, you know, figuring out, you know, how are we going to actually commercialize this idea? And I think what's really taken off is the ease of configurability, ease of use, graphical kind of, for lack of a better word, programming, I'll say. and making this something that the end user can not only use, but also modify or maybe design their own workflows in without having to know a bunch of special stuff. So sure, we provide programming APIs and all this other kind of stuff, but I think we learned along the way that that's not always going to be the most attractive to the customer.

38:58Roby Lynn: Our Sheet Metal customer, for example, they said to us, we don't want to write any code at all. You know, we want everything to be graphical. That's not always the case, but that's what they told us. That was a really interesting, interesting thing to learn. You know, I mean, coming from super tech heavy world we're technical founders, right? I mean, that's the way we think. Yeah. So that was, that was a good lesson to learn, but anyway, so going back to, Oh yeah, go ahead. I was going to start. I'm sort of curious while you're on this thread of, of things you've learned along the way, you know, coming into this, the decades old world of manufacturing as a modern software engineer, right?

39:38You came out of school with all those skills that you just described. What have you seen? Again, PLCs aren't going anywhere. They're deeply embedded into the manufacturing world. And yet the other side of the equation is the advantage of PLCs is that ladder logic is well understood by a certain generation of manufacturing engineers and technicians today. But again, many of those folks are retiring. The workforce is changing and there is a talent crunch. Talk to us a little bit about, from your perspective, what your customers are telling you, what you're seeing about this evolution from the legacy of PLCs and ladder logic to modern software tools, as you just described, and how you see that trajectory going forward.

40:34Yeah, that's an interesting question because I think that kind of feeds into what I was saying our original goal and idea was, we almost didn't understand why ladder logic

40:49Roby Lynn: was needed in the beginning. And it was kind of like, oh, this doesn't make sense. Why not do it another way? We came to learn that there's a reason why it's there because, okay, technicians can debug it simply. Right. A lot of people know it. A lot of equipment uses it and is going to continue using it for a long time. So, like you said, PLCs are not going anywhere. PLCs do their job and they do it very, very well. So I think we had to. It is true that there is a talent shortage. People are retiring. I can't predict the future. Who knows? Maybe one day ladder logic is going to be confined to the history books.

41:29Roby Lynn: Honestly, I don't know. I mean, I was certainly convinced of that at one point in my career, but I'm not totally convinced that that is actually true. Ultimately, I think that it's going to have a place because of those reasons and because a lot more people are coming out of school knowing how to write software now. That's true, but not everybody. So I think software-defined automation, software-defined robotics, these types of modern software for automation tasks is going to become more popular, more attractive. I don't think that it is necessarily going to completely replace everything. I kind of think about this in like, I'm big on transportation history.

42:11Roby Lynn: Okay, I like to think about like how technology changes, specifically how transportation technology has changed over decades and centuries. And like look at trains, for example. OK, we don't use long distance trains really so much anymore to go across the country. I mean, sure, if you want to take Amtrak to California, you can do that. But mostly people are flying. Right. But that doesn't mean that trains have been eliminated. You know, ships have not been eliminated simply because we have airplanes. Things kind of morph and change into where they make sense to be used. And I could go into a whole conversation about I have similar thoughts around AI about this.

42:46Roby Lynn: Right. So my point is, I don't think that. PLCs, they've been proven and they do a great job. You know, I think that they're going to continue to exist in some capacity, possibly, you know, for quite a long time. But I think that their role maybe is going to change. Narrow software is going to expand and there's going to be some kind of combination of the two of those things. And that's where we need this tool that is able to basically work together with everything, if that makes sense. Yeah. Well, one of the things PLCs do really well is the determinism, right? AI doesn't give us that determinism yet that we can guarantee that if we put an AI control on something that it's always going to function the way we want.

43:36And then troubleshooting that, of course, could be a nightmare with the black box where PLCs give us that. And so I think that's one use case or one example for why PLCs will be around for a long time is that deterministic aspect. But you also talk about AI now filling in the gaps, as you said, putting AI for perception, for inspection, those types of tools that you can layer on top of your rack controller. So take us a little bit more about that now, some of the features that your autonomy controller enables to fill in the interstitial pieces to add new functionality where the PLC ends and is not capable today.

44:18Roby Lynn: Sure, sure. So there's a lot of discussion around where does AI make sense in manufacturing? Is it using LLMs to read and interpret SOPs, manuals, debug problems? Okay, there's people working on that, right? Is it coding tools, using LLMs to create code for things? We have a lot of demand and we've seen a lot of interest in just honestly pretty proven AI for vision inspection and object detection, defect detection, counting, types of things that a human can do very easily by eye, but is maybe not so easy to do with a traditional rule-based vision system, especially not in a less structured environment.

45:08Roby Lynn: So in the case of, I'll use the sheet metal one, you know, again, as an example. In this particular case, the conveyor, this is not an optical table. This is not a really controlled lighting environment. because it's not something where everything appears in a predictable way every time and you can create definitive rules that say, okay, measure these pixels and do this and detect this edge and that kind of thing. It's a much more kind of fluid process where we have to train a model in order to detect things. So I think that's where AI really excels right now in manufacturing is doing vision-based tasks, determining if things are good or bad, or if they're passing checks, if the assembly has been done properly, that type of stuff.

45:53Roby Lynn: These models, so these are not LLMs, obviously, right? These models have been around for a while. They work well. They do require, you know, training and inference and all this other kind of stuff that any AI system would have or have to do. And so our platform is designed to natively run those models because, you know, It's GPU accelerated, so we can natively run those models on our system, and also we can train them on our system. So this is another thing I think that is unique here is that we offer training capability directly on the device instead of using a cloud component. And that's attractive because a lot of times manufacturers don't want to connect their OT environment to the Internet.

46:38Roby Lynn: Obviously, they don't want their data getting out. But we've gotten a lot of pushback on subscriptions for things. So they don't want to pay monthly for something that is basically an asset installed on their floor. If they're going to buy it, they're going to buy it. You know what I mean? They don't want to. Why would we rent this when we can buy it? So that, I mean, I think coming from the customer's perspective, that makes a whole lot of sense. They're used to just capital expenditures for buying equipment. And that is what they respond to. Well, that is what they like. when you tell them, oh, you can't CapEx this.

47:11Roby Lynn: You have to pay for it per month. They're not too happy about that. So that's another place where we excel there is giving them what they're used to as a capital asset. But this capital asset is something that can add a lot of intelligence that you wouldn't have been able to do before. Talk about the architecture briefly with the R2 Autonomy Controller, where it sits in the stack, where you would install that. to connect to the PLCs or the rest of the, maybe example of a manufacturing line? Yep, sure. So it's an industrial PC. The hardware looks like a regular industrial PC. This can go right into a control cabin.

47:53Roby Lynn: So just like you would with any other industrial control component. And then we have GPU on board. We have a number of Ethernet interfaces, com ports, kind of stuff that you would normally expect, right? So we can connect to field bus stuff, Ethernet IP, EtherCAT, et cetera. We can connect to IT systems via just normal Ethernet connection. We can connect to serial devices with comp ports. We can connect to remote I.O. systems. So it looks like an industrial controller goes right in the cabinet. Now, obviously, when it comes to the IT-OT crossover thing, there's differences in customers in terms of how they want to do that with their IT systems.

48:39Roby Lynn: And so frequently, there's some kind of integration that has to be done there because maybe customer A has their network set up this way, customer B has it that way. But the point is that everything kind of run through this one box and then all of our software just runs in there and we give a web UI for programming configuration operation. So the customer logs in. Is it talking to the robot controllers too? So if you've got some robots or some linear axes on the line doing assembly tasks, is it also talking to those as a supervisory system? Yeah, yeah, exactly like that. Yeah. And, you know, you could have motion controllers.

49:22Roby Lynn: We work a lot with Bosch Rexroth, for example. You know, you can talk directly to control XOS devices and we can connect Ethernet cameras directly to our system. So gigabit vision cameras. We do OPC way, obviously. So we can talk to PLCs that way. So, you know, the gamut of industrial protocols is really what enables all of this to happen. Thank God for standardization, you know, because that's what really, really makes everything possible. And you're not actually controlling robot motion or linear axis motion with this stuff. You're sitting on top of those controllers that have those tasks of closing the loop and managing process.

50:05You're sitting above this.

50:07Roby Lynn: Yeah, totally, totally. So, like, let's imagine you have a system with, you know, some kind of linear production line with a few PLCs. You have a robot or two, maybe some cameras. maybe you have some MES instructions you want to handle. All of those integration tasks can be run through the RAC. The robot controller controls the motion of the robot. So we could tell the robot go to this position, but we're not going to be doing multi-axis motion control. Multi-axis motion controllers exist, no reason to create another one. We're not going to be telling the PLC to respond to this IO point in some deterministic amount of time or handle safety or whatever.

50:48Roby Lynn: The PLC stays responsible for what the PLC does. The robot stays responsible for what the robot does. But here is a platform that gives you the capability to integrate all those things very seamlessly with web interface. You can interact with everything all at once, collect the data, connect to your MES, ERP, whatever that is. Good. Yeah, I think that's maybe a great description of it. Now, the other product you have is this autonomy monitor. So, yeah, that that is something that is basically the same idea, but without the control piece. So some customers are interested only in maybe we want to monitor a process.

51:25Roby Lynn: I would say mostly what we've found is that. People want to do actions on their data, right? So if you're going to go through and install new industrial hardware in your cabinet, then frequently, as we've seen, as we see the projects evolve, you know, maybe the customer starts out saying, oh, we just want to analyze this thing, you know, and then we'll figure out if we want to analyze this. Almost always it ends up being we want to analyze this and then we want to automatically do X, Y, Z, you know. So we've really started to kind of merge these into one system, which is the overarching rack technology, which is the system that can do both.

52:14Roby Lynn: If you choose to monitor only and not take actions on that, that's okay. But we like to give the flexibility to be able to do both of those things. Obviously, you've started and you go to market here at first commercialization phase in traditional manufacturing, as we just talked about. There's lots of opportunity for deployment there. Are you seeing other markets beyond manufacturing, like warehousing, agriculture, biotech, or food processing, those type of markets as well? Yeah, I think logistics is another big one. That's definitely another big one for us, which is anything from running AMRs to interacting with facility controls and that kind of thing.

53:08Roby Lynn: And this is a similar issue that we see in manufacturing where there's a lot of disparate assets in an environment that need to be orchestrated and they need to work together. So in the logistics case, obviously, we have AMRs, we have roll-up doors, we have elevators, we have security systems, all that kind of stuff. These systems talk to each other in different ways. Maybe they don't talk to each other at all. Maybe an integrator came in and set up a warehouse management system for this, but it doesn't talk to these AMRs, you know, that kind of thing. So we've noticed that there's like similar issues in a lot of places where assets exist, they need to work together and they need to share data with each other and they need to be orchestrated, we'll say, you know, in order to have an entire workflow work well.

53:54So I'll give

53:55Roby Lynn: an example of another customer we have right now. They have a fleet of various types of AMRs, material handling, cleaning, et cetera. They have an existing facility with a fleet of different roll-up doors that all work in a different way. Some of them slide, some of them roll up, right? I should say a fleet of different automated doors. They have an MES implementation. They have an ERP implementation. They have robots. They have controls for their production equipment. But they have a vision of how do we create automated workflows that allow all these assets to work well together. As a simple example, right?

54:34Roby Lynn: So the machine finishes a batch, AMR picks it up, takes it to another place. It has to go through a door. Maybe it has to go upstairs. It has to go on the elevator. It has to do reporting to MES to tell MES where this batch has gone, what machine it's in now, if there were quality issues, et cetera, et cetera. So kind of the ultimate automated factory view, how can we control all of these things together, make them work well as one cohesive workflow? So this one's been interesting for us because, you know, again, same concepts here, but like the different AMRs aren't necessarily going to all work with one fleet manager, for instance.

55:17Roby Lynn: And so we had to be able to talk to the different AMRs, the doors, the elevators, you know, maybe different systems handle those differently. Maybe this door needs to be talked to in a different way than that door needs to be talked to. The PLCs, okay, we have different PLCs on these different pieces of equipment. We had to talk to all of those differently. How do you bring all that under one roof and have one interface where the customer can go in and say, I want to define this workflow to move this product? They don't care about all the technical details of how it happens. All they care about is running their production and running it well.

55:51Roby Lynn: I want to take this product, move it through this process, and have it be automated. it's very similar to the um idea with the sheet metal the painting the defects just different application you know but it's still the same idea it sounds to me after hearing your introduction here to system in the in the company is the integrators the integration network sounds like one of your target customers for your system because integrators are faced with integrating things that are disparate, different technologies and managing legacy systems throughout the years now to be able to come back and upgrade, add new functionality to what they already are supporting.

56:32Roby Lynn: Yeah, that is definitely our ultimate target. We want to give them the tool to make their life easy. You know, you don't have to go in and do all this crazy custom stuff every single time for, you know, all these different applications. here's something that kind of, you know, maybe greases the wheels of that, if you will, allows you to do your job more easily. So that's definitely our ultimate target. We have a direct sales portion of our business also. So I would say at this moment, we kind of work in both camps. We do work with integrators and they will integrate for us, but we also work directly with customers.

57:10Roby Lynn: And it's obviously important for us as a small company to work directly with customers because we get that voice of the customer feedback immediately. We get more visibility for our company. You know, we learn a lot doing integration ourselves, too. So I think as time goes on, direct sales is going to shrink. Integrator sales or channel partnerships is going to increase. But, yeah, it's interesting to think about. I mean, even the big, really big players, even they maintain direct sales themselves, too, which is actually something I learned in this process. I didn't realize that Beckhoff, for example, they do some integration themselves, which was a surprise to me.

57:53Roby Lynn: So I think we're definitely going to – I mean, we enjoy that stuff. I think we're definitely going to maintain that for the long term. Yeah, certainly. Hey, before I let you go, I always love asking this question with entrepreneurs who are learning the business. So do you have any advice for any other entrepreneurs that might be listening, folks that are in your same sort of stage of growth or still at the back of the napkin trying to pull a business together, things you've learned during your tenure as a startup executive? Yeah, let me see if I can figure out how to put this into words. So as an engineer starting a business, I had to learn that it's not always the coolest thing that wins, which sounds ridiculous and obvious in hindsight.

58:39Roby Lynn: But we like cool, complicated stuff. The more complicated, the better sometimes. You know what I mean? And I think we were blinded by that a little bit in the beginning. I think we didn't, we were so set on like, here's something that can be, you know, revolutionary. Why should we pay attention to the existing solutions? That was a mistake, right? And that's not something that, it's a lesson we had to learn. You know what I mean? Like the existing solutions exist for a reason. You look at, you know, a lot of kind of media narratives around the tech industry is kind of like the disruptive technology that makes everything else go away.

59:21Roby Lynn: It's not really the way the world works. You know what I mean? Like there are new technologies that are created all the time. That doesn't mean that old technologies for the most part are just completely eliminated. So I think it's important to be, you know, pragmatic about that type of stuff and not get too wrapped up in the tech marketing narrative of how businesses are built. You know, there's obviously that story sells well to people who aren't entrepreneurs and maybe aren't engineers or building businesses or whatever. But, you know, that's not really the way things actually go mostly. So, yeah, I think talk to your customer, you know, figure out who your customer is talking and ask them what they want before thinking that you know that something is disruptive and therefore everyone's going to buy it.

1:00:08Well, as you said, owning the customer relationship early in the process is important to be able to close the loop with your development team and make decisions about what features are important, which aren't, and where to invest your limited resources until you can grow and get bigger. And as you said, ultimately outsource the things that aren't core anymore to your business. So I will add I will add one more anecdote to this, which is someone on our team one time said, you know, we're building this product.

1:00:41Roby Lynn: The product is going to be so good. It's going to sell itself. We don't need to do any sales work. And which is which is hilariously laughable, obviously. But also, I think that is very emblematic of the way that engineers think about things, which is like we see cool technical stuff and we buy it. Very few people think that way. And I think the tendency for, look, I'm a technical person. Okay, I have a PhD in control systems. I have been an engineer my entire career. I had to come to the realization that the world does not revolve around engineers. You know, there's lots of other equally important roles in a business.

1:01:21Roby Lynn: And one of those is sales and another one is marketing. And the business will not survive unless all of those things are working well. So my advice to any engineer would be, don't think that you're the center of the universe because like everybody has their role, right? Well, Roby, it's been a great conversation. I want to thank you for coming on today and sharing your business with R2 Labs and what you guys are doing. Absolutely. I wish you all luck and looking forward to following your growth in the coming years. Thanks so much, Mike. Appreciate it.

1:02:02Awesome stuff there. Obviously, we talked about this a lot, right? One of the hardest problems in industrial environments is connecting all these disconnected systems, Mike, right? All the different vendors, all the different workflows, trying to bring that together into one coherent process. It seems like a tall order. I think we've heard from a lot of folks that it's chaos trying to bring these older systems together that have ultimately been patched work over the years. But I like how we seize the opportunity to try to allow this integration to happen all in one place through software. So kudos to Roby and the R2 Lab team.

1:02:37But what were your takeaways from that conversation? Yeah, I agree. I mean, I learned so much in that conversation with Roby. And, you know, really the challenge that exists is, you know, for folks that work in the manufacturing space, the classic integration scenario where, you know, trying to support product lines that are doing the same thing, building product, packaging things. I've been doing it for more than a decade. Stuff just, as you said, just gets patchwork built along the way. as well as the skill set to support some of those tooling, PLC programming, all the rest, retiring and trying to bring people in to support legacy systems.

1:03:20So that's really what Roby sees as the opportunity. And I loved, as he said, they have learned so much as they have peeled the layers of the onion away, so to speak, of the manufacturing legacy to be able to understand where the opportunity is to leverage modern AI tools in the OT space, to fill in the gaps, to understand maintenance needs, to predict maintenance needs, to program and reprogram systems on the fly to meet the continued need for improvement in quality and all the rest. Being able to throw in now vision to do quality checks at different points and to quickly integrate that into the whole process to add new capabilities as well.

1:04:11Another important opportunity. So, again, kudos to Roby and team. Again, it's another company we will continue to follow their innovation and growth in the coming years. Yeah, it's always fun to see them. So again, they're going to be at Robo Business in the startup area, booth 723. If you're looking to connect with R2 Labs or Roby, Rebecca and the whole R2 Labs team, I'm sure will be there. Rebecca is one of the best persons. I don't know if you've ever seen her, if you've ever run into her. I forget if it was our show or another show, but she was doing push-ups. She like got people to do a challenge.

1:04:45It was later in the day at whatever show it was, and she was doing a push-up challenge to get the blood flowing. So maybe we can have a fun, see who can beat Rebecca at a pushup contest. Maybe we'll have an adjuvant humanoid that's going to be there. We'll put them, we'll pin them against each other and see who can do. I know she's listening. So she's definitely tenacious. So she's a great asset for their team. You know, she kept, she kept hounding me. You got to get Roby on the podcast. So here it is. Yeah. Awesome stuff. So again, before we go, just one final reminder about Robo business, commercial development, commercial robotics development event coming up October 20th and 21st out in Santa Clara, California.

1:05:24This is the 20th anniversary show. So we have some fun stuff planned. The agenda is live on the website. We have a few things that we'll be adding here in the coming weeks and months, but go check it out. I think it's going to give you a great feel for what you're going to learn about and who you'll hear from at the show. We'll have over 70 people on our stages over the course of the two days talking all things commercial robotics development and humanoids and AI and scaling robots and deploying robots, field robots, all sorts of things. But again, we're trying to bring together people who have actually deployed a large number of robots in the field over the last number of years and have them share their war stories and all the challenges and all the triumphs that they've gone through to try to lift up this industry.

1:06:08So again, if you want to come to the show, if you want to get involved, head over to robobusiness.com to learn more, you can reach out to any one of us here at the Robot Report. Our early bird pricing ends at the end of the month. So if you're coming and you want to get a full conference pass, you'll save a couple hundred bucks if you register within the next 11 or 12 days. So make sure you do that. But again, I hope to see you all at RoboBusiness. This has been a very important show to the robotics industry over the years, certainly for myself, Mike, and for you and Eugene. So we put a lot of work into these shows.

1:06:41So we really hope that we see you guys there at RoboBusiness out in Santa Clara. But that's going to do it. Episode 258 of the show. Thanks again to everybody for tuning in. Thanks to Roby Lynn, founder and CEO of R2 Labs for joining us on the show today. New episodes of the Robot Report podcast drop each week. You can find us anywhere you listen to your podcast. Please subscribe. Leave us a rating. Give us a review. This episode was produced and edited by Mike Oitzman. Double duty this week. Mike, thank you. Our executive producer is Amanda Buhner. I'm Steve Crow. Thanks so much, folks, for tuning in.

1:07:15We'll talk to you again next week.