“I felt the pain of being an integrator and envisioned a world where an entire workflow could be entirely digitalized.”

Picture building an automation system in hours, just like with LEGOs. This is what Vention set out to do when Vention began in 2016. In this episode of Manufacturing Matters, Etienne Lacroix, Founder and CEO of Vention joined TECH B2B Marketing’s Jimmy Carroll to talk about how Vention aims to break down the barriers of entry when it comes to automation system adoption. He also compares what Apple did for music – with the integration of iTunes and the iPod – to putting automation hardware and software in one simplistic, plug-and-play ecosystem that is more easily accessible. Additional topics include automation growth drivers, AI and machine vision trends and benefits, building an ecosystem through partnerships, and more.

Jimmy Carroll: [00:00:07] Hello, everybody. My name is Jimmy Carroll. I’m the vice president of operations at Tech B2B Marketing. And welcome to this episode of the Manufacturing Matters podcast, where we aim to discuss the trends and technologies reshaping the manufacturing industry today. I have the pleasure today to be joined by Etienne Lacroix, who is the CEO and founder of Vention. Etienne, thank you so much for taking the time. I really appreciate it.

 

Etienne Lacroix: [00:00:27] Thanks, Jimmy, for having me. Quite excited to be with you here on the podcast.

 

Jimmy Carroll: [00:00:31] Well thank you. Yeah the feeling is likewise, obviously. So for those who don’t know, can you tell us a bit about Vention and what led to you founding it.

 

Etienne Lacroix: [00:00:38] Yeah, Vention is a story that started in 2016, where we got the idea to democratize industrial automation by giving the tool to the practitioner, the folks on the factory floor. And what we set ourselves to do is to create that very Apple-esque business model, taking hardware, industrial automation hardware, and software, all the software you need to design and program those robot cells and automated equipment, and bringing them into the same plug-and-play ecosystem. And the intent here was to create a very simple, very delightful user experience so more people could adopt industrial automation. The ideas actually came from myself, and younger in my career I was a system integrator, working out of Montreal. We were mostly serving the automotive and the aerospace industries, and I’ve navigated probably 40 to 50 projects during those years. And what I would define as the traditional fashion, right? You go on an industrial distributor website, you find parts, you open SolidWorks, you design additional custom parts. Eventually you have a machine. Then you have to open Studio 5000. You start to program the PLCs. Eventually all those things come together. But it was very project-based, right? And as a result, one of two projects, we were literally losing money as integrators. And it was a pretty hard business. You have amazingly talented engineers, but a very, very hard business model because obviously you bid for winning those jobs. You take sometimes a little bit too much risk, and one out of two times you’re not making as much money as you thought. So I felt the pain of being an integrator. And around that time, 2016, you could envision a world where that whole workflow could be entirely digitalized. If you have LEGO parts to build a machine with. So that’s a little bit the founding stories and how we came up with the concept.

 

Jimmy Carroll: [00:02:34] Yeah. So that’s really interesting. There’s a lot of follow-up questions that I want to ask you, and I will. But before we dive into that and the Vention platform and that LEGO idea, which I love, a little further, I want to ask about, what are the pain points people are facing today? So manufacturers, logistics facilities, fulfillment centers, whatever, places where automation is being adopted. What are the challenges people are facing today, including the labor shortage? This is an issue that’s not going away. Obviously it was sort of highlighted and COVID exacerbated it to some extent. But it’s still here today after all these years. What are the pain points you’re seeing for these companies today? And how does industrial automation technology fit into that equation?

 

Etienne Lacroix: [00:03:18] I think as a continent, in America in general, we just don’t automate enough, right? You look at the level of adoption of robotics in the United States versus Germany, for example, or even Korea. We are so little versus what’s being done over there, and there’s such room to automate. And I think one of the main reasons people don’t automate as much is because it’s so difficult and it’s so pricey. If you’re a high-throughput manufacturer, you can buy automation on a project basis and still have very, very good ROI, and you have enough volume to absorb all the initial CapEx and all the expertise that needs to come with the integrators to put those systems together. But if you’re a small-medium business, and you have a very small production volume, like sub-10,000 units a year, it’s starting to be pretty hard to pay for all those technologies, the hardware, the software, the integration services, and absorb all that CapEx with small volume. And that’s why people don’t automate. That’s why people rely on laborers, which is, to your point, Jimmy, very difficult in today’s world because there is labor shortages, and it’s not going to get any better. As we all know, the world is getting a little bit smaller, right? That’s going to create additional pressure on finding laborers to be in our factories and operate those manufacturing floors. And so those problems are going to become even more exacerbated. And that’s why automation is going to be vital. And to make automation really something that can be adopted at the same pace as what mobile phones were adopted or PCs were adopted, you just need to make it so much simpler.

 

Etienne Lacroix: [00:04:53] And, as I said, we decided to take kind of the Apple approach, and bear with me for a second. I think this analogy is worthwhile for the listeners. When I was a kid or a teenager, music was free, right? You can get the entire Radiohead discography in two minutes if you were willing to navigate torrent and burning CDs and all that stuff. And today everybody pays $1 per song. Back then, my parents could not consume digital music. Yet today, everybody can. And what Apple did was a magic piece of genius, is they basically decided to come up with the iPod and the iTunes store, hardware and software, working together as one to create a super-simplistic user experience. And it tilted an entire industry that was basically free, to paying along the way, enabling everybody to adopt digital music. So how about we do the same thing for industrial automation, right? Hardware and software intended to work together as one in a simplistic ecosystem. So everyone on the shop floor is able to automate. One of the things I like to say, and it’s an aspirational saying, is, like, it would be great if the folks that today are operating the floor become the folks that tomorrow are designing the floor, right? And we’re not fully there yet, but I think we’re definitely working towards that vision.

 

Jimmy Carroll: [00:06:12] So really a fantastic answer. A lot for me to ask and follow up. I love that you gave an analogy, also one of my favorite bands, so I can appreciate that very much. I want to ask, though, you mentioned small to medium-sized businesses. And I think that’s a question I probably don’t ask enough on this podcast, but industrial automation technology, in terms of adoption, the numbers according to some of the big outfits, like A3 or Interactive Analysis or some of these others, they’re overall mostly positive. There’s maybe a bit of flat growth this year or last year because the numbers went up precipitously after COVID. Those numbers were always going to come down a little bit. But for the numbers to continue to grow, for automation to continue to grow at significant numbers, percentages, whatever, I’ve always felt and I’ve seen it echoed in the industry that if that’s going to happen, then it needs to become more cost-effective and easier for small and medium-sized businesses to be able to adopt the automation, because these big companies are already using it for the most part. So do you agree with that? Am I right or am I off-base?

 

Etienne Lacroix: [00:07:29] I agree with your point. Industrial automation for high-throughput factories has been solved. There’s great offering. The technology is mature. It works well. What hasn’t been solved is how to go after small-medium businesses that have the exact same need, right? They’re facing a labor shortage. They’re feeling cost pressure. Yet there’s not a good offering for them. And I think it’d be simplicity, simplicity, simplicity. So more people can do them by themselves. And there’s a couple of trends that are helping with that right now. We like to talk about platforming and productization. That was our approach. Like if you develop this very big ecosystem of plug-and-play components and you know that everything that is available is compatible, it brings the barriers of adoption lower. That was one way. Productization to me is just as interesting. Today, it’s very visible that in the industry there’s productized robot sales, whether it’s a palletizer, a sending cell, a welding cell, now they’re productized kits, and they are sold usually with a performance level. And that removes all integration almost from those projects, making them more like a product and less like a project. It makes those projects less risky. And that’s probably what’s remaining. If you’re a small-medium business, you have a need, you find one of those productized applications. The risk profile is much more interesting. But if you go to something a little bit more custom, something that has not yet been productized, that’s where you’re back in the traditional way of doing industrial automation, where you need a lot of integration expertise to bring those solutions. So I think we’ve solved the technology risk portion by platforming and productization, removing technology risk. What we haven’t fully solved yet is process risk, right? When you automate a task, there’s a lot of process knowledge intricacies that system integrator have developed over the years of banging their head at solving given problems. And now they have all that expertise, and if you need to put a certain type of object into a box, they know exactly how to orient the box. And that knowledge has a lot of value, but it’s really hard to do that today in a productized fashion. So I think the next frontier is around how can we reduce process risk? All that knowledge that was developed over the years around the manufacturing process can be perhaps embedded in a software or in a platform. And I think AI, for example, is going to help quite a bit with that.

 

Jimmy Carroll: [00:09:59] Yeah. I mean, you mentioned welding, for example. I know that it’s a serious concern that the majority of skilled workers in the welding industry, at least in the United States, will be retiring at some point soon. But the idea of putting that expertise and a specialized product together into one solution that can solve that problem in the future, sort of getting out ahead of that, and there’s a lot of different examples like that, but I think welding is a really good one. As far as the idea of making it more simple, you mentioned LEGOs earlier, and my oldest son is 8. He’s a LEGO master. And we’ve got a lot of LEGOs in the house. I step on them every day. But I love the idea. Again, my brain works well with the analogies of sort of building industrial automation systems like you build with LEGOs. So tell me a little bit more about that if you don’t mind.

 

Etienne Lacroix: [00:10:58] I think just like your your son, Jimmy, I was a big LEGO fanatic when I was a kid myself at the time. LEGO Technic, the one where you have suspensions and gears and some motors, and I did probably so many of them from 8 to 10 years old, up to probably my early 20s, where I played with those, those LEGO Technic. And so the concept of modularity was quite understood. In fact, when you get later on in your LEGO career, you think about what you’re going to build before you open the LEGO box, because all the building blocks are stamped in your brain. You know exactly: I’m going to build a four-wheel suspension car today, and you envision the entire design in your head before you open the LEGO box. That concept is very, very powerful because you can do the design off-line. And equipped with that concept and that pattern recognition and equipped with the understanding of where CAD technology was going back in 2016, we knew that if we were to marry the two, something magical would happen. And in 2016, that’s where WebGL — WebGL is a little library that we used to do 3D in the browser.

 

Etienne Lacroix: [00:12:08] And today it’s ubiquitous. But before that point, you could not do engineering-grade 3D in the browser. It was too CPU- and GPU-consuming. You had to do that on a desktop, install software. And if you bring that experience on the desktop, you’re no longer connected. The environment where you design cannot be the environment where you buy, right? That has to take place in the browser. And that was possible since 2016. That’s where we envisioned, like if we put all those LEGO parts, industrial LEGO parts, robot arms, conveyors, motors, sensors, structure, and we put that in a cloud-based CAD software that understands all the various parts can be connected with one another, you can do so much automation within the design and lower the barriers of adoption. Today, simple stuff like just generating a bill of material, automatically, as you design; computing the assembly time, managing all the fasteners for the various parts that we always forget at the end of the project and then you’re losing a couple of days; managing; recommending the best part. There’s tons of parts to search and find when you’re doing those machines. What if the software could recommend the best part and learn from everything else that was designed before? When you start to work with a LEGO part inside design software, you can automate a lot of tasks, and those tasks basically just make the software simpler to use.

 

Etienne Lacroix: [00:13:30] And that’s very important because now you can give that software to folks that usually have very good understanding of the manufacturing process. Perhaps they’ve never learned SolidWorks or CATIA or Inventor, but if you reduce the barriers of complexity, you make it very simple, they know exactly what to do, right? And that was the experiment we were up to back in 2016. And it worked beautifully. Today there’s around 2,000 LEGO parts, I would say, in the Vention ecosystem, and we keep adding more. But to make sure we don’t create complexity with all those LEGO parts, we’re going to look at additional features like AI-based part recommendation or sentence completion, but applied to 3D design. So when you mouse over a given mechanical context, will snap right in place the part like you can go there, a little bit like when you’re in Gmail and you’re writing a sentence, the next few words are appearing. We can now do that in the context of 3D design. So all of this is really geared to make it very, very simple so more people can adopt.

 

Jimmy Carroll: [00:14:34] Yeah, again, I love the analogies. It helps me picture it as a non-engineer. So if somebody is building one of these, if they’re speccing out a system and using your platform, when they get to the end and they’ve got a system that hypothetically will work together, what happens then? Like, for example, first, what do they do next? And then does Vention also do systems integration services or is that something you shy away from?

 

Etienne Lacroix: [00:15:01] So the story I just described is true. But there are some folks that will need support as well. So what we see is there are groups of users, manufacturing professionals, they want to do everything by themselves. They want to design by themselves. They want to deploy by themselves. Do the production ramp-up by themselves. And that class of users do exist. And that was one of our big questions when we started Vention. Now, as the machine gets more and more complex, you will see a class of user that either need design help or deployment help. Usually the design help will come first. And there’s ways to get connected to an application engineer and you get advice. And usually they will need just a couple of advice here and there and off they go and they want to be left alone. Now a lot of clients want to have also delivery services or deployment services. And usually this is very correlated to the complexity, again, of the application. And today we do offer system integration services within our ecosystem. So we have a team that’s sole mission is to go at client, deploy solution, start the machine, train the team there, make sure that the clients get autonomous, and get out. And we’ve done that all across the United States and in Europe. And if you think about, as we’re building Vention in this ecosystem, this pattern is actually the same that a lot of companies have done throughout their career.

 

Etienne Lacroix: [00:16:28] If you remember, when FANUC started in the late ’70s, they actually had their own system integration in-house services. They’ve learned how to crack various manufacturing problems using the robot they came up with, and eventually they built an amazing network of integrators across the United States. Same thing was true for Boeing. I’ve learned that just a few weeks ago. But Boeing, when they were building their first new aircraft in the ’20s, were also operating their own airlines. And eventually they pulled out of the services layer and enabled a network of airlines to flourish. I was talking with the founders recently doing electrical vertical takeoff vehicles, and they were going the same route, like we’re building those electrical aircraft, but we’re also going to operate them as flying taxis, because we want to learn all the intricacies and the business model properly before we can give that. So to some extent we’re repeating that pattern as well. So this is a very proven way to bring a new way to go to market, new ways for people to automate, and making sure that everybody stays happy and satisfied and obviously come back as clients.

 

Jimmy Carroll: [00:17:39] Yeah. Again, there’s a lot there. I want to ask a lot of follow-up questions, but I told you upfront I’d be respectful of your time. So there’s a couple of trend questions I wanted to ask you about. And you mentioned a little bit earlier, Vention uses AI-based product recommendations. And I want to know a little bit more about that. But also I wanted to get your idea of AI in general. Just a minute ago, you said the word “ubiquitous” – AI is ubiquitous. It means different things to different people though. In the mainstream, some people get afraid of AI. They don’t quite understand what it might do, not realizing that it’s in their phone. You use it every day. You just don’t really realize it. What does it mean to you? What does AI mean to Vention and its customers? And what are some ways that you’re seeing AI add value to industrial processes today, both within the context of your customers and the general industry?

 

Etienne Lacroix: [00:18:35] It’s such a rich topic. AI can be used to increase the ceiling or the performance of an automation solution or AI could be used to lower the floor or improve the adoption. We’re using AI to lower the floor and make it more accessible to more people. I’ll talk about a couple of use cases where we’re bringing AI in a second, but the general trajectory is that us and the entire industry is up to right now is to bring to bear autonomous robotics applications. And by that I mean robotics applications that just need much less user input to perform the task they’re meant to do. And for us, AI is very unique because — because Vention is cloud-based and because over the years a lot of people have made designs on the Vention platform, and all those designs are made with LEGO parts, that means there’s a rich dataset that we can learn from, and we’re using that dataset to predict the next best part, for example. So those prediction algorithms today get informed. And as people are playing with new parts or testing new components, we can know if a component is better suited in a certain situation. And we’re leveraging that to make the bar of design simpler.

 

Etienne Lacroix: [00:19:52] But you stay mostly in the mechanical context. Now when you move to the automate context, the automation side of things, Vention on the platform as a programming environment, there’s one that is code-free, there’s one that is programming Python. With that comes a copilot. So you can generate prompt, right? You can say, “Hey, give me a function to index the conveyors by 100 mm every time the robot completes a cycle.” You can just prompt that in text. And the function will be generated in Python. And you can put that in your program if you want to. You might have to edit two or three things here and there, but you’re there 80% or 90% of the way. Now the beautiful thing is, when you program a machine in the context of its digital twin, and all the digital twin is fully labeled. There’s a conveyor, there’s a robot, there’s an apple in the bin that needs to be put on the conveyor. Like you can start to integrate those parts from the digital twin into your prompt and get a program that no one is referring to. So it becomes very, very powerful. And again, you lower the barriers to adoption. Now let me get to the last piece of AI, and I think that’s the one that is the most exciting but the one that’s going to take the most time, which is getting to autonomous robotics applications. And for that, that means you need to have a machine learning model behind every robot deployed on the floor. We’re not there yet. That also means that there’s vision on every robot deployed on the floor. So today, the way people do AI, or physical AI they call it, is they’re going to buy all the expensive equipment, the robot, the camera, the GPUs, and all that. They’re going to start to do a task, learn from it, and eventually the machine learning model will be good enough that it can handle all the edge cases and pick the part in all various situations. And the problem with that is you need to front-load all the CapEx to know if the application is going to work in the end. And if you’re a small-medium business, that just doesn’t work for you, because you want to make sure it’s going to be risk-free, right? So if you can take all the training, the model, and virtualize it, do that in the cloud instead of doing that on the edge, that means you can de-risk before you commit to the CapEx investment.

 

Etienne Lacroix: [00:22:05] So you need a virtual camera. You need virtual grippers, a virtual digital twin, where you can actually train the model, test it, like if you were in real life, before you commit to CapEx investment. And with this you can start to build those autonomous applications and a much more risk-free profile. I think that’s what’s exciting about it. I think we’re going to see — academic examples I like to call them. We see them already. We’re going to see more in the next years. I think we’re probably at two to three years away, where a sizable amount of applications – by sizable I don’t mean 50%, but I mean like 5%–10% of the applications deployed leverage some sort of AI on the shop floor. And if we use that AI to make robot programming significantly easier, I don’t have to program an approach point and a retract point, and you can just go grab the apple, and the robot can figure it out by itself and avoid all collision and figure out the fastest path planning. So we can do that at scale. That would be great. And that means we’re closer to folks that operate the floor becoming the folks that design the floor.

 

Jimmy Carroll: [00:23:09] Yeah, yeah. AI is, like I said early on, like it means a lot of different things and it’s applied in different ways. And based on my conversations, I’m not on the factory floor. I’m not actually doing these things. But I talk to a lot of people who are, and some of these things are further along than others, right? Like at first I thought generative AI might have no place at all on the factory floor and industrial automation, and to some extent it does. And in other places, it’s still getting there. Like programming, PLC programming or robot programming, might be something that’s really valuable. I know big companies are doing that now. I think Beckhoff and Rockwell and some others, Siemens, but I did a podcast a while ago with a gentleman who was director of AI for a systems integrator. And he said, one of the things you have to worry about with generative AI, as much as I see the potential value for certain applications, is it’s maybe not there yet, and he provided this example, which I thought was really interesting, and I’m going to steal it again. He said, “Hey, tell me the names of all 12 of Snow White’s dwarves,” right? Obviously there’s only seven, but the model just spit out 12 names, and the additional five names sounded very familiar, very similar to the other seven names. So I give the ChatGPT or whatever it was credit for coming up with something similar. But obviously these AI hallucinations can’t take place in an environment where such a mistake could be really bad. So it’s a topic that’s very interesting to me. So I always like to ask about it.

 

Etienne Lacroix: [00:24:51] AI is a non-prescriptive technology. We’re used to dealing with very prescriptive technology. You put an input, you get an output, and regardless, if you put the same input, you will always get the same output. And AI is non-deterministic. So not prescriptive. Me and you could put the same prompt in ChatGPT and we’ll get different answers. So that’s why it’s a non-deterministic technology. And in the world of manufacturing, where we’re dealing with precision, like submillimeters, like precision, it can get scary, to your earlier point. What if for the last 4,000 picks you got it right, but then the next one you hit the collision that the robot did not plan. And here we go. Production is down for three days. And so I think next year we’ll see more academic examples. But there’s still a couple of years for us to prove the technology and make it to a state that is ready for the manufacturing floor.

 

Jimmy Carroll: [00:25:47] I could ask a lot of other follow-up questions on AI, but I’ll save that for another time. I did want to ask, something that’s curious to me, Vention’s a unique company in what you guys do, so I imagine that partnerships there are very important. I mean, they’re very important for everybody in the world of industrial automation today. But you’ve got a couple of exciting ones, a couple new ones. Can you talk to me about these and maybe tell us what this means for your current or potential end users?

 

Etienne Lacroix: [00:26:17] So building an ecosystem is something you need to do with people. With partners. So this is not something we ever envisioned being able to do on our own. So we’ve been partnering with Universal Robots and FANUC for the longest time. And I think those relationships have been fantastic for them and fantastic for us and fantastic for our users, bringing that simplicity and that proven industrial technology together. There’s two big partnerships we’ve announced this year. And the way usually we announce partnership is when we’re ready and when there’s a product. So we’ve been a little bit conservative on that side. And the first one we announced this year is ABB. And our story with ABB that joined the ecosystem and the GoFa family, their collaborative robot family, is now fully integrated in the Vention ecosystem. And ABB is a fantastic relationship for Vention and for our users because we used to partner with them since probably 2021 in Europe, where the team in ABB were using Vention to complete their own robot cell when they were serving their their client. And so they’ve been to some extent a client of Vention. And so we were working with them on that basis. And what happened more recently is we started to work with ABB also here in the United States to automate factories in other divisions.

 

Etienne Lacroix: [00:27:31] So they’re using Vention technologies, so they became also a client from that point of view. But we’re helping them on a very strategic mission of automating their factories. So we got a little bit closer and at the beginning of last year, 2024, we had a discussion between the various groups, and it now was the time to bring the GoFa family. We had so many, two or three years worth of proven relationship working together, and it was time to bring the GoFa. So today we bring the GoFa in the Vention ecosystem. That means users can not only design with the GoFa; they can program and simulate with physics or in code-free or in Python with the GoFa. They can deploy them through the Vention stack and get all the simplicity that we provide with the analytic suites, the remote support suites, and the teleoperation suite that Vention provides. So it’s great because we are again lowering the bar or lowering the ceiling, the floor for people to adopt great cobot technology. So we’re feeling extremely proud and privileged to be working with ABB. The other partnership we announced this year and was also one in the making for over a year was Nvidia, that is now partnering with Vention. Vention and Nvidia have been working together on various aspects but mostly in the context of our plug-and-play motion controller machine. MachineMotion. Today, for the last years we had generation one of MachineMotion. And today we’re providing our client with generation two, MachineMotion v2. And we’ve announced a little bit earlier in the fall, MachineMotion AI, which is our third generation. MachineMotion AI is what I would define as one of the first products of the post-PLC era. In a world where those robots will have machine learning models behind every single one of them, you’re going to need GPUs. And right now, the best GPUs that you can get for industrial technology come from Nvidia. So it’s very natural for us to partner with them. To give you a sense, the whole thesis behind MachineMotion is it’s actually pretty hard to do an industrial automation cabinet. You need drives. You need PLCs. You need safety PLCs. You need power supply. And you need very qualified electricians to wire everything up. And you create one of those. It takes a few weeks to serve a single machine.

 

Etienne Lacroix: [00:29:53] And we went back to that thesis of LEGO-like and productization. And can we create a box that fits it all up? That is always good. So that’s how we created MachineMotion, which contains everything that I just described, but in a productized fashion. And we wanted to remove the burden of requiring the electrician. So everything is plug-and-play with connector bays. So MachineMotion AI pushed the frontiers a little bit more, where you can do up to 30 daisy chain motors fully synchronized through Ethercat, 3000 watts of power, and I/O links ready, IP54 enclosure with passive cooling, over-the-air upgrade, because a lot of plants have, for very valid reasons of cybersecurity considerations, so you want to be able to have an autonomous machine not being connected to the plant IT. And obviously there’s GPU in there, enabling to run 2D, 3D perception models directly on the controller. So now you need one box. Everything is plug-and-play, and you can run all the future of automation without needing any additional devices. So that was the thesis behind it. Very happy to to work with Nvidia on that first product. And as you can expect, Jimmy, there will be other products coming on in 2025 as a result of that great collaboration.

 

Jimmy Carroll: [00:31:11] Yeah, I mean it’s super-exciting. Both ABB and Nvidia, I mean, anything Nvidia always catches my eye, over the last couple of years, just a fascinating company. And one of the things you said to me offline, when we previously spoke, was, “I don’t understand why anyone would need to use a PLC once they can see what this product can do.” And I said, “Oh, wow. All right. Well, I’m gonna have to ask you about that.” It’s very interesting. One of the follow-up questions that I wanted to ask you about was, you mentioned support for 2D, 3D cameras, but does Vention also venture into — I guess I didn’t do that on purpose, “Vention venture into” — machine vision? Can users design that into the process on your platform? And what’s that look like once the system is deployed?

 

Etienne Lacroix: [00:31:58] Very good question. So MachineMotion AI is vision-ready. We support PoE cameras. We support cameras for teleoperation as well. And our pendant, the HMI that is connected to MachineMotion, also includes cameras. So when you call remote support, think of this as telemedicine but for your robot cell. We see you, we see the machine, and so on. So we’re ready from an infrastructure perspective. You’ll see, throughout the year in 2025, us venturing into releasing some of those vision-based applications. But really the trajectory here is one of walking towards autonomous robotics. And so you’ll see those announcements throughout next year.

 

Jimmy Carroll: [00:32:44] I hate to bring up too many things that we spoke about offline, but they were really interesting, so I just want to ask about it. One thing that you said, again, was related to machine vision. And I geek out a little bit with machine vision because in a previous life I was a writer for a machine vision magazine. You told me that you think that we’re about three to five years away from every robot cell having a camera. So I just wanted to bring that up.

 

Etienne Lacroix: [00:33:14] Well, yes, but I’m still conservative, right? There’s a lot of problems that can be solved with just good mechanical design. And every time you can solve a manufacturing problem with good mechanical design, I would always recommend, let’s just solve it with a good frame. A good machine frame, good positioning, good referencing. You make the life of everybody else in the chain so much easier. But then the cost of a camera is getting so cheap, right? And they will become so ubiquitous, that at some point I wouldn’t be surprised if some of the engineers, I don’t want to say get sloppy, but you will not need to be as skilled in geometric referencing and positioning and machine frame alignment, because you’ll always be able to compensate with vision, right? You can have a crooked machine, but it doesn’t matter if the parts are not presented properly. The robot will figure out what the parts are. And so that future to me is not that far away, right? And I wish we don’t get sloppy too quickly and we keep things simple with just good mechanical design. But I think there will be a point where the cost of doing those designs right and the expertise needed versus the cost of just putting a camera on and the software being pretty autonomous to figure it out, the trade-off will happen. And, yes, perhaps that trade-off is between three and five years. I’ll give you an example we mentioned. We only have three motor sizes: small, medium, and large. And that trade-off basically implies that the time it will take you to right-size the motor is not worth the pricing difference between each of them. So think about the expertise needed for electrical engineers to right-size based on the load, the peak load, the power consuming, all that. You could spend a lot of time just right-sizing the motor on the machine. And, we said, well, if you spend three or four hours for a system integrator, that’s several hundred dollars, and you’d rather just go small, medium, large, and pick the one that is most likely. You’ll end up with a slightly over-designed machine, but it’s just not worth the cost. And I think the same trade-off will happen with vision. The cost of doing the right mechanical design with all the expertise needed – let’s just add a camera and leverage the machine learning model, right? We’ll be in that trade-off in maybe three to five years.

 

Jimmy Carroll: [00:35:29] In some cases, to your point, there are a lot of applications that don’t require vision, but there are certain applications, and I’m basically looking for your opinion on this, but it’s sort of an opinion I’ve formed over the years, adding cameras, oftentimes 3D but 2D as well, adds a new layer of flexibility. And then to your point, adding machine learning, AI, deep learning capabilities adds even further level of flexibility, again to your point, to be able to let’s say inspect a part that’s maybe off skew or whatever.

 

Etienne Lacroix: [00:36:02] You add robustness to those applications and a fail-safe. Like we’ll deploy the machine and if we need one more lever to make sure it works when we deploy it, well, the camera’s already there as part of the cost. It’s pretty low cost and now add a camera. And yeah, it gives you that additional robustness, that additional fail-safe to make sure that when you deploy, you will be able to continue to improve the performance of the machine. And, yeah, definitely there’s a future where vision still can be tricky at times and you still have to fight with lighting conditions and all that. But we’re slowly, slowly getting there.

 

Jimmy Carroll: [00:36:34] I’m glad you mentioned fighting with lighting conditions. I’ve got some some friends in the industry that are vehemently opposed to the idea that AI can compensate for ambient or poor lighting, so they’ll appreciate that.

 

Etienne Lacroix: [00:36:49] There’s still a little piece of technology to fix for sure.

 

Jimmy Carroll: [00:36:54] Fair enough. Etienne, what else haven’t we talked about today that you might want to bring up? Or have we covered a good, wide variety of topics?

 

Etienne Lacroix: [00:37:03] I think it’s just an exciting time to be in manufacturing. I got super-passionate about manufacturing. I was in my very early 20s, and at the time I was probably the only one passionate about manufacturing and industrial tech. All my friends wanted to go into tech startup and stuff like that. And I think now it’s actually cool again to be in manufacturing. There’s so much technology coming up, right? There’s a lot of people that can contribute and be and be part of this. So I think for the industry as a whole, it’s quite exciting. And we’re reaching a point where the technology that is available to us is there to really make a dent in adoption. To really make adoption change versus the previous years. So I think we’re going to have a very exciting 10 to 15 years ahead of us here in this industry.

 

Jimmy Carroll: [00:37:52] It’s something that I’ve been part of for 10 or 12 years now, but even in those 10 or 12 years, a term that I think I’ve seen you use or your company anyway is the democratization of automation. I think about whatever the technology may be, 3D or hyperspectral. Hyperspectral, for example, was a technology that was limited to the lab, and it was very large and expensive and sort of inaccessible. And now, or SWIR, like SWIR imaging now, the cost is way driven down. It’s available from a number of leading machine vision camera companies now at affordable costs. And same for 3D and so many other technologies. And your platform is a reflection of a lot of different converging technologies. And that’s why I wanted to talk to you today. I reached out and said, I want to learn more about Vention, and I’ve been seeing you guys around at trade shows and, yeah, it’s been really interesting. I really appreciate your time. For folks that want to learn more it’s Vention.IO, or if you have questions for Etienne, reach out to us at manufacturing-matters.com. I’d be more than happy to pass along those questions, and I hope everyone enjoyed it. And Etienne, one more time, thank you very much. I appreciate it.

 

Etienne Lacroix: [00:39:15] Thanks, Jimmy, for having me today. It’s very kind of you.

 

Jimmy Carroll: [00:39:19] Of course. My pleasure.