Speaker 0 00:00:00 Customers, leverage player's expertise in machine vision and AI to digitize a process as they scale their way towards full automation of a process.
Speaker 2 00:00:18 Hello everyone and welcome to the Robot Industry Podcast. My guest for this edition is Ed Goffin from Liora Technologies, Ed's the marketing manager and their experts in ai, machine vision and smart manufacturing, and they're based in Ottawa, Ontario. Before we get started, I wanted to welcome Canova. Ken Nova's, a global leader in professional robotics, and they're founded in 2006 in Montreal. The company's mission was initially to empower individuals with upper body limitations through use of assistive robotics. The company has evolved its product line to service researchers, medical professionals, governments, business and educational institutions to achieve their innovation goals. Through strategic partnerships today with robotic technologies built up over more than a decade, canova is dedicated to provide solutions to professionals in, in industries, from AgriFood, healthcare and security to nuclear, hazmat and advanced manufacturing. And we welcome Canova as a sponsor. Ed, welcome to the podcast.
Speaker 0 00:01:18 Thanks for having me. It's a pleasure to join.
Speaker 2 00:01:21 Ed, can you tell us a little bit about Pleura and, and what do you do at Pleura? And I, I, I, we're gonna talk a lot, talk a lot I today about vision and about, uh, AI and about combining those elements.
Speaker 0 00:01:32 Sure. So, uh, Playa, we're actually a business that's been in, uh, existence for 23 years. We celebrated, I think our 20th year, right in the middle of Covid. For the vast majority of our 23 years, we've been focusing on the machine vision market. We sell components that go into cameras, uh, industrial automation systems, medical imaging, security and defense applications. And then over the last couple years, we started looking at some of the expertise that we had, you know, around connectivity, realtime video delivery, standard standards based expertise so that you can integrate different types of cameras and processing. We're like, okay, where else can we go with this expertise? At the same time, we've started to have more manufacturers come to us asking about, you know, how can I use automation in my domain? I'm interested in in ai? Can you teach me more about ai? Uh, and so it led us into developing a suite of solutions more geared towards an end customer manufacturer. Uh, myself, I'm marketing manager at Playa. I've been here almost 10 years before joining Playa, I've been in the telecom industry, various types of networking applications that all basically look like vision and now increasingly today incorporate vision into those types of applications. So my worlds are coming together. Uh, in terms of background,
Speaker 2 00:02:59 You're published and you've been a speaker, uh, on topics from packet switching to network timing to real time video solutions for kind of that situational awareness that you just mentioned, but in medical factory automation and in security. So, uh, I've glad that you were able to make time today, uh, prior to the automate show, and I'm sure we'll talk to you a little bit about that as well.
Speaker 0 00:03:20 Yeah, you caught me just as I was finishing up my automate presentation really, so it was perfect timing.
Speaker 2 00:03:25 So you've said before that people want more automation and they want to use ai. They got, they say, I want ai, but then they say, what is AI again? So what are some of your thoughts on that?
Speaker 0 00:03:37 Yeah, usually that's how most of our conversations start with the manufacturers. Uh, they, they come to us looking for a solution to a problem. Oftentimes believe that AI is the solution to the problem. Part of my joke is I, you know, marketing is in my job title and I blame marketing a little bit for some of the hype around ai, but we usually spend the first part of that conversation really learning, you know, what are you trying to achieve, right? What's the problem you're trying to solve? Uh, what are the tools that you're using now in terms of cameras and machine vision applications? And is AI a fit? You know, some of those applications are inline applications. A fair number of those applications are actually a process where a manufacturer's using a human to do the process and they would like to, their goal is to automate that process. And usually as we talk them through, um, their requirements and, you know, their appetite for cost and complexity, we usually start them down a path more of digitizing that process first with the eye of heading towards automation.
Speaker 2 00:04:44 So what are those first steps in digitizing and, and using some of these technologies?
Speaker 0 00:04:49 Yeah, you know, I can give you a couple of examples. There's, we work with an electronics assembly operator and they have, um, they work in, uh, a custom market where they'll make a few hundred boards for a customer that's going into some sort of aerospace application. They have full scale automation for a lot of their products, but for these smaller runs, it's a little bit more challenging to automate the process. It's more expensive. They came to us initially looking at, you know, can I find a way to automate this? Um, and as we talked through the discussion, it was like, no, let's, let's see if we can first digitize this. You know, give, use some basic image, compare tools from machine vision, give the operator really guided assistance around looking for and finding errors. And behind the scenes there's some unique things that we can do around training and AI model so that they can gradually work their way towards full automation.
Speaker 2 00:05:46 Who are your bullseye customers? Are they the small and mediums, or are they the large guys, or are they integrators?
Speaker 0 00:05:52 It's a little mix of both. I mean, for the most part I would say it's the small and medium. Uh, they're the ones, especially if they're looking for a solution to a manual process, they're the ones that probably rely the most on a manual process in their manufacturing facility. Now, saying that, I'm always surprised when I get to go into a facility and, you know, we've been into some pretty large plants, just how much humans are still actually involved in manufacturing processes. We'll see next week at automate we'll be wowed by all the robots doing all sorts of things. And that happens in a factory, but I think it's something like 70% of processes still rely on a human. So saying that, you know, the small to medium are definitely the target for us, I would say. Um, it's where we have the best conversations when we get into the larger manufacturers. They're, you know, they're looking at us more for AI algorithm and machine vision algorithm types of solutions. So we serve both markets, but definitely have a, a really good fit for those small and medium manufacturers looking to digitize an automated process.
Speaker 2 00:07:04 One of the things that we mentioned in the, uh, uh, just before we we turned on the record button is you'd mentioned that your S D K is probably in a lot of places that we'd never know and some o other background technologies. Can you expand on that a little bit more?
Speaker 0 00:07:20 Yeah, so that goes back to our machine vision, uh, roots. Um, we have I an S D K called ebus, the most widely used s d SDK in the machine vision world. And a lot of camera manufacturers rebranded as their own sdk. So there's, if you're using a machine vision camera in an application, there's a pretty good chance that player's SDK is in that camera in some fashion. The other area that's interesting is around medical imaging. So if you've, if you've ever had an x-ray, you know, a broken bone or a dental x-ray, I can almost guarantee that player's connectivity solution, uh, got the images from the x-ray panel to processing. It's, uh, we're designed into pretty much every x-ray panel out there that's in medical applications.
Speaker 2 00:08:08 And that's really exciting. Of course, that goes to a lot of the technologies that you've developed over the 23 years. Um, I wanted to switch the conversation to talk a little bit about the return on investment of automation, robotics and AI from your perspective.
Speaker 0 00:08:22 Yeah, that's a great question cuz again, that, that comes up pretty early in conversations with customers, right? Um, I'll say we usually get involved in projects sort of in two key areas. One is, and back to one of your earlier questions, you know, they're excited about AI and they wanna learn about AI and they've sort of followed this path around a deployment option usually, you know, based on best use case scenario. And then they go to deploy it and they pretty quickly run into, you know, this is a little bit more complex than a best use case scenario. My defects are unique, my products are unique. My, you know, I'm talking about a short run production and this use case was geared towards a long run production. Um, this is more complex and taking me longer than I thought. Or the other big one that stands out is my employees are resisting this change usually, you know, as, because it hasn't been properly communicated to the employees what you're trying to do.
Speaker 0 00:09:27 So that's where, you know, we get involved sometimes with the customer where they've reached that disappointment stage and we can sort of step in and say, okay, let's, let's simplify this project and digitize instead of automate, um, you know, take a first step. We can gather data as we digitize with the plan of automating, but let's digitize this troublesome process you have on your production floor. And usually they see a pretty fast return on investment. I mean, we're working with one customer now, they do, uh, again, it's in the electronic space and there's work instructions around a machine set up and it takes 'em on average an hour to set up the machine. And then they digitized that process and gave the operator work instructions and it became a five minute process.
Speaker 2 00:10:19 Well, that's a great example. And I guess educating your clients because of now the interest in ai, that is a big part of your job.
Speaker 0 00:10:27 Yeah, I mean that's what really for the last, you know, two years, um, it's mostly what I've been doing, you know, going to a lot of events, talking about some of the customers that we're working with. We've had a couple of really interesting use cases around electronics and a distillery that uses, uh, some of our digitization tools. Uh, and just talking about what were the thought processes of those types of customers as they look to automate and, and what are some of the quick wins they can get with a digitized first, automate second type of approach.
Speaker 2 00:11:02 And so do you get into a lot of situations where you're helping the customer in their process?
Speaker 0 00:11:09 Yeah, so usually what happens is our, our salespeople have a real solution focus, uh, and want to understand the problem that a customer has. Could be a, a telephone call, it might be a site visit, but really explain to us your problem and show us as best as possible, show us the actual defects. I mean, that's what I love about these trade shows and I'm sure we'll see it at automate. I know I saw it last year a lot or manufacturers walking around the trade show with a defective part, right? And saying, can you guys fix this for me? I remember seeing like, you know, textile, it was a, a person, they had a bag full of different types of textile and they were getting streaks across this textile, I think it was like a vinyl flooring. And we're like, can you solve this for me? So usually those are the types of conversations that we, we like to have. And I mean, those are the conversations that you have to have, I think, to, to be a proper partner with your customers.
Speaker 2 00:12:06 And there's lots of humans, as you mentioned in these processes, and you don't wanna remove the, the people, right?
Speaker 0 00:12:13 Yeah, I mean, there's a lot of cases where the manufacturer doesn't want to remove the people for various reasons. I mean, the, the one that is most obvious is it's usually a, a smaller production run, uh, usually a very high value product and usually in low volume. So to automate that can be too expensive, right? If you're looking at full scale inline inspection and especially, or if you're a regional product, right? You're switching around products a lot, um, you know, different labeling for different regions and that sort of thing, it gets expensive to set up automated inspection and those types of applications. So there it makes sense to keep the human. The only problem with humans is that we, we get tired and we get distracted and we start to make mistakes. Usually, usually actually the mistake we make, which is sort of an interesting one, is we start to see errors where errors don't exist. So we start to see false positives and end up stopping production. And a lot of people then, uh, end up gathering and looking at products and trying to decide is this a good product or a bad product?
Speaker 2 00:13:23 No, that's a really good point. I hadn't thought about that false positive issue. Very, very, uh, interesting because that one more time when people shut down the line and they're all looking for something that doesn't exist.
Speaker 0 00:13:33 Yeah, that's exactly, so this distillery that we work with, that was, uh, their, um, CTOs or quality manager's main issue was they had a, they have a labeling issue and they use a mix of automation and manual processes to put labels on bottles and to package the bottles. And it's, if the labels don't line up, you notice it mostly when it gets on the store shelves. So for us in, in Ontario, they send their bottles to the L C B O, and as soon as the stock or at the L C B O starts to notice that the labels are askew, says, no, I can't have this on my shelf and sends it back. So then that puts a lot of pressure back to the employees of, okay, we've gotta get this right. They were finding that if, if an employee had any shred of doubt, they would stop production. Then you'd end up with 20 people from across the manufacturing floor all gathered around looking at labels on bottles and trying to decide is it good, is it bad? People see things different, there'd be arguments on the shop floor, this is good, this is bad. And he's like, I'd I'd like to find a way that this is consistent and subjective all the time.
Speaker 2 00:14:43 No, I think that's a great, uh, example. And of course with your technology, there's no crooked labels, right? Or it gets shut down really quickly.
Speaker 0 00:14:51 Yeah, they can see, they use the tool in a couple of different ways. One is an inpro guidance for the operator. So it shows them where on the bottle the label should go. And then there's a really fast quality control check where again, it's a camera based system. They can pass the ca the product under the bottle, the bottle under the camera manually, and it quickly tells 'em, pass fail. You know, this is within intolerance, the labeling is all right. And this is where you can get into some AI training too, right? It can start to look at are my labels of good enough quality? You know, the young cousin runoff, they do regional products, so do I have the right label on the right product for the right market? You know, all those things that we as humans, we can make mistakes or overlook with a trained model, it, it starts to spot those differences right away.
Speaker 2 00:15:40 No, that's a great example. Do you have any other use cases that you can mention?
Speaker 0 00:15:44 Yeah, there's, uh, you know, there's a couple of other ones in sort of more the, uh, metal parts. Uh, you know, they want to look at threading on screws. One is, is there enough lock tight on the screw? If I put, um, you know, a couple of rings of lock tight on, it's perfect. If I go to three or four then it's imperfect. Uh, so they're, again, they're doing it manually cuz they're not doing, you know, batches of thousands of these products. It's a few hundred. So they want to manually look, we get brought into a lot more products, uh, problems around electronics inspection. So is the assembly correct, uh, as it's coming off a machine? And then is the assembly correct as the operator's doing the next steps on that product? Right. Usually they're putting together something, so have they used the right screws? Have they not damaged any components as they're doing the second step on the board?
Speaker 2 00:16:44 That's great. And um, where do you think that vision and AI is going?
Speaker 0 00:16:50 Yeah, we were saying before we, we hit record there, there're worlds that are definitely colliding. You know, the machine vision market itself, I think up until a couple of years ago, it was a fairly insular market, right? There was, you know, cameras and cables and processing and, you know, we served a select group of customers, um, and the market was doing fine. If you look at market data around the machine vision market, the last couple of years, it's exploded mostly because our products are now getting pulled into all these applications that we probably never thought about five years ago in terms of some of its AI inspections, some of it's like we mentioned before, security and defense around local situational awareness. And even in those markets, again, it's security and defense is an interesting one. It's, can I give a crew real time video in my vehicle from all these different sensors connected on the vehicle?
Speaker 0 00:17:49 Next stage is, okay, can I start to also use some AI in these applications so that I can start to spot things that my crew commander has to be able to find right away. It, it's just a, an interesting collision and I think we'll see that again at Automate. I mean, automated Self has changed as a show. I know we went 10 years ago, we hadn't been for a couple years. We went back last year and I was like, man, this show has really changed a lot because vision has pulled into all these different types of applications that I hadn't even thought of.
Speaker 2 00:18:20 And are you seeing, like I the from some of the use cases you're mentioning it's about quality in some other places it's other things. Are you seeing any trends?
Speaker 0 00:18:29 Yeah, the, I mean, quality's the big one. We, we ran a survey of a number of manufacturers and we're just compiling the results now and sort of asked what is, what's your priority generally, number one. But number two was around, um, returns. And we've got a few customers that are working on projects in that space too. It can be expensive if you're making a lower volume, higher quality, higher price product. Finding yourself in a situation where you're dealing with a customer around an infield fault, usually those conversations are long, you know, we're learning from manufacturers, sometimes they just throw out their hands and say, fine, we will take back the defective part even though I don't think I made the mistake and send you a new part. And it's expensive. Uh, so they want to try to get into a stage where they're digitizing sort of that product shipment process and capturing more images of the product as it's not only going through sort of inspection stages, but also the packaging and the shipping so that when they have these R m A conversations, they have some record of, okay, this is what this product looked like when it left my facility.
Speaker 0 00:19:38 So that was number two, was around product returns. How do I, how can I protect myself, uh, against these expensive and usually lengthy conversations.
Speaker 2 00:19:49 Thanks for that, ed. That's very interesting. And did we forget to talk about anything?
Speaker 0 00:19:53 No, I mean we talked about a fair bit there. <laugh>, it's an interesting world, right? Because, um, again, like I said earlier, we, we sort of start our conversations usually with a lot of manufacturers where they, they want to automate a process. I've been working with them to take a step back a little bit and say, okay, you know, automation is our end goal and are there ways that we can get there? If I digitize a process for you, reduce the errors, you can get a return on investment right away. And also I can start to gather some of the data that then as you take that next step towards automation makes it a little easier. For example, let's say the distillery would like to eventually automate inspection by using that q qc check. Mm-hmm. <affirmative>, they're actually either just able to start gathering images that they'll be able to then use or will be able to use for them to start to create those vision and AI models as they move into inline inspection.
Speaker 2 00:20:46 Yeah, that's a great example. So, uh, let's talk about automate show. Uh, do you know your booth number
Speaker 0 00:20:52 <laugh>? I've, I've written it enough the last couple of weeks that I do know it off the ha top of my heart. It's one 18. Uh, you know, as you enter the facility you go, right, it was sort of funny last year. Last year you might remember we were sort of still in that slightly covid rule thing, so only one entrance was open, right? And it was, um, right in front of our booth. So every morning from like nine to noon just was an onslaught of people. So I see this year they've, they've opened up more doors, so it should be, our days should be spread out a little bit more
Speaker 2 00:21:25 And what we have in your booth, um, and, and you'll be chatting about AI and, and digitization and such.
Speaker 0 00:21:31 Yeah, so we will have, uh, we have three demos that really cover the three main product lines for player. So on one side we'll have a solution that we called Vera that is a, an app-based platform for manual inspection applications. So it, it's the solution I've been talking about where you can digitize a process so you can digitize and automate visual inspection and add AI assistance for your operator. Uh, so it'll be showing a live demo of that tool. We'll have a live demo, a solution we call ebus Studio, which is an AI machine vision algorithm development platform. So it would just be running a small conveyor belt, uh, and using AI to identify shapes and sizes of metal parts running over there and showing an AI confidence model and triggering a PLC when there's a fail, just sort of showing the intent of that platform.
Speaker 0 00:22:28 It's a low code platform that you don't necessarily need to be an integrator to design an algorithm. We give you the tools so that you can start to design and deploy your own AI algorithm. And then we'll have, uh, a third demo where we're partnering up with one of our customers lab for, or a Canadian camera manufacturer based on a Waterloo. They use our ebus sdk and uh, they're looking at partnering with us down the road around, uh, a smart camera core concept. So this is again, back to more of our machine vision roots, but we'll be expanding our S D K platform to add some capabilities so that you can design more flexible AI based cameras.
Speaker 2 00:23:12 Oh, that's very exciting. And, um, how can people get ahold of you?
Speaker 0 00:23:16 Probably the best way is LinkedIn. If you look, uh, look up my LinkedIn PPL profile, ed Goffin at Playa or, uh, email is fine too. Ed period goffin playa.com
Speaker 2 00:23:28 And Playa is spelled p l e o r a.
Speaker 0 00:23:32 That's right. <laugh>.
Speaker 2 00:23:33 And when you're not changing the world through AI and vision and automation, what do you like to do? Do you have any hobbies?
Speaker 0 00:23:39 I'm a big music fan, so I love going to see live shows, uh, and collecting music and listening to music. Usually, usually most of, I'll say my vacation, but vacation for my wife and I, <laugh> will include a live show somewhere trying to see a band at a cool spot or a, a venue that I've always wanted to go to. So, so yeah, when I'm not working, I'm usually listening to music and sometimes I'm listening to music while I'm walking our dog.
Speaker 2 00:24:08 Nice. Thanks for, uh, participating today.
Speaker 0 00:24:10 Yeah, thanks for having me.
Speaker 2 00:24:11 Our sponsor for this episode is Airhart Automation Systems. Airhart builds and commission's turnkey solutions for their worldwide client. With over 80 years of precision manufacturing, they understand the complex world of robotics, automated manufacturing and project management, delivering world-class custom automation on time and on budget. Contact one of their sales engineers to see what Earhart can build for you and their
[email protected]. And Earhart is spelled E H R H A R D T, and I'd like to acknowledge a three, the Association for Advancing Automation. They're the leading trade association for robotics, vision and imaging motion control and motors and the industrial artificial intelligence technologies. Visit automate.org to learn more. Painted robot builds and integrates digital solutions. They're a web development firm that offers SEO and digital social marketing and can set up and connect c r m and other e r p tools to unify marketing, sales, and operations. And if you'd like to get in touch with us at the Robot Industry Podcast, you can fi find me, Jim Beretta on LinkedIn. We'll see you next time. Thanks for listening. Be safe out there. Today's podcast was produced by Customer Attraction Industrial Marketing, and I'd like to thank Chris Gray for the music, Jeffrey Bremner for audio production. My business partner Janet and our sponsors, Airhart Automation Systems, and Canova Robotics.
