Visions Podcast: Cognex Talks AI Vision on the Factory Floor

This episode of Visions: A Machine Vision and Automation Solutions Podcast, explores Cognex's recent survey on AI-driven machine vision, featuring an interview of Cognex CEO Matt Moschner by Sharon Spielman, VSD's head of content. The discussion covers years of AI tool development, common adoption hurdles, usability and scalability improvements, and industry-specific adoption trends in automotive, electronics, logistics, FMCG, and semiconductors.

Listeners will learn how modern AI vision advances are reducing deployment time, enabling lift-and-shift across global sites, and where adoption is progressing fastest.

Related: Unlocking AI-Driven Machine Vision: Insights from Cognex's CEO

Related: Next Gen Machine Vision: Anomaly Detection for High Speed, High Mix Production Environments

Transcript

Well, hello and welcome to Visions: A Machine Vision and Automation Solutions Podcast. I'm your host, Jim Tatum, senior editor of Vision Systems Design and visions is an Endeavor Business Media production from your friends at Vision Systems Design. Here you'll find the latest on everything from end user machine vision solutions to trends, developments, and perspectives on all things machine vision and imaging. Whether you've been working in the industry for a while or you're just starting to take a closer look at it, this podcast is designed to grow your knowledge and bring greater focus to your understanding of the imaging and machine vision industry. And now on to our show.

AI driven machine vision isn't a concept of tomorrow. It's changing quality and efficiency on factory floors today. Cognex Corporation recently surveyed more than five hundred manufacturers about how artificial intelligence is transforming machine vision systems in industrial production. The report, which was just released today, reveals shifting priorities, rapid adoption and evolving expectations across industries. I'm Sharon Spielman with Vision Systems Design and joining me to discuss the real-world impacts, challenges, and future directions for AI driven machine vision is Matt Moessner, the CEO of Cognex. Welcome, Matt.

Hi, Sharon. Thanks for having me.

So I was able to look over the survey that you are just releasing and shows significant adoption of AI and machine vision, especially in automotive, electronics and logistics. So from your perspectives, what are the main factors driving this momentum? 

Yeah, absolutely. I mean, I would, um, I would say we've been working with these technologies for almost ten years. You know, we made really a hallmark acquisition in late 2016 of systems, which brought to us deep learning tools that had been specifically targeting industrial inspections. We made a second acquisition in 2019, Korean company called Su Lab, which really gave us not just technology but critical scale in terms of talent and engineering momentum. And, you know, we are today on what you might say is our fourth generation of AI vision tool development. And I'm happy to talk about where we've been focusing across those nine or ten years and those multiple generations of technology. I think you rightly point out we've seen customers really adopting this technology in bigger and bigger ways. And that was really what underpinned the survey that we ran, right? We were seeing just really increased demand from our customers for the technology, and we wanted to better understand more analytically, frankly, what they were looking for, what they care most about, and what maybe was limiting their adoption more completely. And, and I think we uncovered some really interesting insights from that survey. 

What were the initial hurdles that manufacturers commonly face when adopting the AI based vision systems? And, of course, Matt, how do they overcome them? 

Yeah. Of course. Well, you can imagine in the early days it really was, will it work? And, and I think we've overcome that, right? The performance of today's AI vision tools are tremendous. Um, and they are doing things that, you know, we frankly couldn't imagine even five years ago. And so I think we're past the “will it work” question, but now it's into, can I support it? Can I scale it? Uh, and can I use it broadly? And so that's really a usability question and a scalability question. And that's really where we've been focused, uh, I would say for, for the last several years, which is obviously continuing to take advantage of the technology progression, right? As AI models and more powerful models have been released. We're at the forefront of that but making sure that we're also using the technology to make using a vision system simpler. And so we're bringing that into the product in different ways. You know, with things like auto tuning sequences, we've really almost completely redesigned the workflow of our vision systems to be more data centric than programmatic in their design. And that has given customers really two things. It lowers the barrier to trial, right? So in the first generation, there was quite a bit of upfront engineering investment that our customers would have to make, whether it be image collection, training schemes, validation. We've really, as we've streamlined that, the time it would take from an idea to a pilot is hours, not weeks. And, and that really matters that that improves their ROI. It gives them more confidence to use it more broadly. And so I'd say that's the biggest shift from will it work? Yes, it's going to work, but can I use it? Can I deploy it? Can I scale it? And that's really where we focused. And I think you saw that from the survey results. 

So, for companies that are like operating multiple sites globally, what are the key considerations for that successful scale for these deployments? 

Yeah, I think it's a question of, you often put a lot of investment in that first pilot station or site to say, you know, is this going to solve the problem that I have? And when that's successful, many of our customers are multi-site, multinational, multi-geography producers. And they want to obviously take that to other sites. And so the question is, can I quote unquote lift and shift the work that I did from one to the other? And that's a question of how robust is the technology meaning how tolerant is it to variations in line design potential, you know, environmental factors like lighting, glare. And what we've seen is as the tools have improved their ability to be ported from one environment to the other without really any intervention or change has really dramatically improved. And so think about what that means for a manufacturer, they can have the confidence to say, well, if it works over there, I know it will work over here. And often, as you know, there are slight changes in the product itself based on where it's made. So you might have, let's say, a shampoo bottle manufactured in the US for the US market, which is very similar, but maybe with a slight redesign in packaging markings for the European market produced in Europe. Our tools can make those modifications very quickly without having to re-initiate an entirely new product. And so I think that that really speaks to the scalability and how quickly you can take a pilot to the entire network. 

So were there any industry specific or regional differences in AI adoption or challenges that really stood out in the survey results? 

Yeah, I think you pointed out a few of them, which was, you know, we go to market through five primary market verticals and really three of them logistics, consumer electronics and automotive we're seeing is really leading in terms of adopting these new tools. And that's not surprising. And I think on one hand, these are the areas that stand to benefit the most. Either had problems that some of the older tools and technology couldn't address that the newer can. They tend to have very sensitive ROIs when it comes to even very small improvements in vision, system performance, and detection performance. And so for sure, we're seeing those three verticals and customers in those three industries leading, I would say, in their adoption. And you know, the markets that tend to be a little more cautious would be our fast-moving consumer products. And, you know that has a number of reasons for it, right? These are customers that tend to be very sensitive to, you know, quality are heavily regulated industries because they're selling into consumer end markets. And so we're seeing a little more caution there. And then, you know, our fifth would be advanced semiconductors. And similarly, I think, you know, that's an industry that tends to be a little more cautious in its adoption of leading technology, given the criticality of the machines and the expense of the manufacturing process there. But we are seeing nice progress as well. You know, in terms of geographic adoption, I think it's really quite balanced from our customers in Asia to the Americas, to Europe. I wouldn't say we saw any meaningful trends geographically. 

Well, that's a wrap for this episode of visions produced by Endeavor Business Media, a division of endeavor B2B. Thanks very much for tuning in. If you enjoyed today's show, be sure to subscribe to the podcast and share this episode with a colleague who would find it helpful. Until our next episode, you can find us at vision dash systems dot com or on LinkedIn, Facebook, or for more insights, updates, and breaking news to keep you in the know. Thanks for tuning in. Until next time, stay focused on your visions.