Episode 39
Joe Cyrek
Magswitch Automation Group

 

On this episode of Manufacturing Matters, Joe Cyrek, President of Magswitch Automation Group joined Winn Hardin and Jimmy Carroll at the 2024 A3 Business Forum to talk about the important role of switchable magnetic technology in applications including robotics and automation, welding, manufacturing, and material handlings. Specific applications discussed included automotive manufacturing, bin picking, electric vehicle battery cell handling, cobots, and even shipbuilding. Additionally, the podcast covered warehouse automation growth drivers, ease of use in automation, and the potential use of AI within switchable magnetic technology.

Magswitch_Logo_1200x1200

MagSwitch-Business Forum 2024 PG.mp3: Audio automatically transcribed by Sonix

MagSwitch-Business Forum 2024 PG.mp3: this mp3 audio file was automatically transcribed by Sonix with the best speech-to-text algorithms. This transcript may contain errors.

Jimmy Carroll:
Hi everybody. My name is Jimmy Carroll. I'm the vice president of operations at Tech B2B marketing. I'm here today at the Manufacturing Matters podcast, at the A3 Business Forum in Orlando. I have the pleasure of being joined by Joe Cyrek, the president of Magswitch Technology, and my colleague Winn Hardin. Joe, thanks so much for taking the time.

Joe:
Thanks for having me. great event. I guess just quick shout out to the A3, Jeff Bernstein, and his whole team for just being consummate professionals and doing so much for the membership here. Really appreciate the opportunity to be here. And thank you guys for inviting me on this beautiful Tuesday morning here in Florida.

Jimmy Carroll:
Absolutely. Yeah. As long as you're indoors. Joe, so for those who don't know, if you could give us maybe a quick primer on what Magswitch does and what you do there, that'd be helpful.

Joe:
Fantastic. Yeah. So I'm the president of the Magswitch Automation Group. So Magswitch has been around since the early 2000s. And we have technology that can turn a permanent magnet on and off. So here in Florida, many of us have the little Mickey Mouse Walt Disney World magnets stuck to our fridge. We have a way to make that not a magnet. Turn it off, turn it on, on a bigger scale. But so we make switchable permanent rare earth magnets that are not electromagnets. Use cases are robotics, automotive, heavy industry, agriculture, railroad, anything that's ferrous steel. We pick things up and put things down on the end of a robot. We're not robot integrators. We're not robot manufacturers. We make the widgets, but they go on the end, and we make widgets that can pick up a paper clip. And we make widgets that can pick up the side of an aircraft carrier.

Winn:
Good Lord. So I'd like to see that one in action. Yeah.

Joe:
It's a big widget.

Winn:
I got to check the YouTube channel. Yeah for sure.

Joe:
So that's us.

Jimmy Carroll:
From an application standpoint, you touched on some of those from a high level, but I want to ask about a few in particular. One is a rise in application-specific systems that come purpose-built for specific applications like bin picking, for example. So you've got a robot that comes out and it's incorporating 3D vision and software, sometimes AI based, and then some sort of robotic end-of-arm tooling. Do you see your products being used in a lot of bin picking or pick-and-place type applications?

Joe:
Uh, 100%. It's actually been our biggest growth segment over the past 24 months. And one of the unique reasons that a Magswitch is so great at bin picking is because it's a single-sided picking device. So if you picture a 4 foot by 4 foot by 4 foot deep bin full of randomized metal parts, an overhead vision system finds the part. It tells the robot where to go get it. But all those parts are just in a mess down there. And so traditional methods, somebody would come in with a two-finger gripper and try and grip it, and maybe the finger would hit the part next to it and then move the part it was going to get. So in our vernacular we call that hugging. So if you're using grippers or clamps, you have to go in and figure out how you're going to hug the part. With a Magswitch, we're a single-sided gripping device. We just hover down, go where the vision system tells the robot to go, and then we turn the magnet on and we like to say we kiss it, and then we extract it from from the bin. So a single-sided gripping device has huge benefits in the bin-picking world for chaotic parts.

Joe:
And we're just one piece of the puzzle, but we think our piece enables a more complete bin pick and enables an easier programatic setup for the robot programmer, for the vision system, and then our tools, some of our tools also have onboard intelligence. So as we go down and we kiss that part, our tool will tell the system that, yeah, hey, you have the right part, that the part is currently all there and present. Sometimes parts have brackets that are added or weld nuts or features. So our tools can tell the difference if we have one part or two parts, if it has the hole in the part that the part is supposed to have, and it has the bracket welded to the part that it's supposed to have, because we can measure the density of metal through the magnetic field and give that information back to the robotic system. In the event that we did pick a wrong one, the robot can then turn the magnet off, drop it, and we can go find a right one or hold it, put it in a reject bin, and then start over again.

Winn:
So just change the orientation of the product in case it's got multiple pieces.

Joe:
So in automotive it's been it's been a very popular option for Magswitch for the past 24 months. And then we're seeing things in the steel industry like steel service centers, forging centers, where after the grinding and polishing, it's in a bin and maybe they need to pick it out of the bin and load it onto a pallet and then shrink-wrap it, and so we can hold that stuff. So that's a little bigger tool. So with the current generation maybe not wanting to come to work so much, the automated bin picking and line-side loading, where whether it's line side or if you have a little farm, the company has a farm that loads all these carts, and automated guided robots take it to the line for delivery. It's going to be a very successful growing field for us and for everybody here at this conference, from the vision guys to the robot integrators. The downside for us, I hate to mention it, but fair is fair. Aluminum we can't do.

Winn:
You need good conductive materials.

Joe:
If I had the aluminum magnet, we'd be doing this on my boat somewhere in the Gulf of Mexico, but I don't have that yet.

Winn:
We got to have something for next year. Joe, can you give any kind of guidance about the specificity, either spatial resolution ability of kissing versus hugging? I mean, obviously you can go small, but it's just in terms of pick accuracy, maybe repeatability. Is there any kind of general guidance you give folks?

Joe:
We tend not to. So the other cool thing about our magnets — a magnet has a north pole and a south pole, right? Our north pole and our south pole are actually added-on features to our magnet housing. And they can be contoured to any shape so we can match geometry of a truck floor pan. We can match geometry of a washing machine support bracket.

Winn:
Kind of combining a fixture with the magnet itself.

Joe:
You got it 100%. And then we can also add a pin in the middle, a non-ferrous pin between the north pole and the south pole that acts as a guide. And so from a vision system or a bin picking, if we get close enough, we sometimes equate magnets to the old "Close enough is good enough," like in a hand grenade. It's horseshoes and hand grenades and Magswitches. You get within a half an inch of your part and you turn the magnet on and like the Star Wars tractor beam, it sucks it right up into your locating pin and into your into your fixture magnet. So huge advantages there that literally can't be done with traditional clamping.

Winn:
What's the financial argument you make? You say, okay, we can combine basically gripper with fixturing systems, whereas normally you might have to do a whole separate mechanical systems. Is that a driver for folks to adopt this technology too or is it just to increase throughput, accuracy.

Joe:
It increases throughputs. So on the bin-picking world, everybody wants to get 99 to 100% bin complete, right? Because if you take a bin away and it's still got 20 parts in it, that's no good because now somebody still has to dump that. So the availability to get 99 to 100% of the parts, and bin picking is very important for us and our customer base. and the customers enjoy that. When we come to fixturing and putting Magswitches in fixtures, whether operators are loading parts, it's actuation time. Our magnets for the most part turn on in 250 milliseconds. So a quarter second on, quarter second off. There's no kinematics. So if you're designing the fixture that's getting loaded, you don't have clamp open, clamp closed. If it's an operator loading the fixture and the safe distance to walk through the light screens to allow the safety, there's no movement, so you can shorten the safe working distance. Shorter light screens, shorter walk path, faster cycle times.

Winn:
Better for retrofits, especially in space-constrained operations, which almost, what 75–80% of your installs probably, if not more.

Joe:
And then the easiest one is if you have a robotic system picking up steel parts and you're over cycle by two seconds. And you have big clamps, and nothing against the clamp people, and it takes a second to close. Switch those clamps to magnets and you just save three-quarters of a second on the open, three-quarters of a second on a close. There's the low-hanging fruit of 1.5 seconds in your cycle time and per robot segment or station.

Jimmy Carroll:
Well, it seems like another value there that you guys might be able to add is you probably don't need vibratory bowls or shake tables as often for small parts, right? You kind of remove that need.

Joe:
It's a great question, and absolutely. So it's two basic schools of thought with our single-sided picking. So again we have this bin of parts, small brackets. Vision system tells the robot to go. The robot goes and gets one somewhere in the center. Next vision cycle comes. The vision systems typically want to find something near their origin. So you're always going into the center. So we're going down, we're picking one, we're picking one, and we're digging a hole. The other school of thought with Magswitches is you go down, you turn the magnet 100% on, and you just got a barrel of monkeys. Your one part that you wanted is stuck to the magnet, but you have all these other ones, and now you know that to keep that one part, I need to turn my magnet on just to 22%. So you turn the magnet down to 22%, and then the barrel of monkeys is flattened out. So it's still holding the one you wanted and you go do the thing you wanted, but now your surface for your bin pick has been redistributed and reorganized, so you're never going to dig a hole in the center. Every time you drop that barrel of monkeys, it flattens out. So instead of, as you pointed out, doing the shaker table or picking the bowl up and swirling it around, you don't need to do that. Save cycle time, save cost. We just reorganized the whole deck of cards for you.

Winn:
The fast switching, and I assume this is part of that intelligence. Is that AI based? Is that a deep learning model?

Joe:
It's not yet. That's in our product development road map. We have the circuitry and the computing power, if you will, inside there. AI is on our road map. We're just in the infancy stages of learning what we don't know. But right now, we can tell you when we pick that part that we have the part, it's the correct part, that the high-low operator didn't load the wrong bin of parts and we just picked one that's a half a mil thicker. So sometimes Model A, Model B, Chevy to Cadillac may have different thicknesses of metal. All our tools can tell those different thicknesses. So we know if this is the Expedition or the Navigator or the Suburban or the Escalade.

Winn:
And most of those times you're not even having to use vision, right? You're just using that through magnetic field sensing?

Joe:
Yeah, it's none with the vision. It's all through our tools and the magnetic field sensing. And so we can do that also on the fixture side, if you're loading fixtures, sometimes parts have to be stacked and there's no way to tell if that part was loaded behind the current one that's there because you can't shoot it with a proximity switch. The vision system can't see it. But our magnetic field detector can determine it. In the stamping world. One sheet, two sheet, three sheet, double blank detection before you load. Nobody wants to load two pieces of metal into a die and hit the go button. Oops, right? We just need to make sure there's always one one sheet in there.

Winn:
So two or three years ago you started hearing about people who were deploying AI grippers. So in general, can you give us some thoughts as an expert in that side? Is this a real thing? You said it's on your road map. You know, neural networks came on the scene 15, 20 years ago, and then AI really came on seven years ago or so. And we're finally starting to find some really good applications for it in volume. How does that fit into the robotic gripper part?

Joe:
You know, for us, the big conundrum or the hardest sticking point is the fact that every piece of steel is different. Like the metallurgy of . . .

Winn:
You're going to get variations, natural variations.

Joe:
It's crazy how many different steels from the steel companies, although they all look the same on a car or on a Bobcat or on a Caterpillar. Yeah, they're completely different. So where we're at right now is trying to train the AI to learn what these different metals look like through the magnet.

Winn:
I honestly thought it would be an easier model, because I'm sitting here thinking, you don't have 100 different defect classes like a machine vision model, but but actually you do.

Joe:
It's way worse, you know. Look at these tripods with the lights and cameras here today. If I put my magnet on this one made by Manufacturer A, and it's going to give me a reading that says, Oh, this is 127, and that one looks the exact same, but it was made by somebody else from a different material. And that one's 142. Now our goal is to have the magnet pick up the part, tell you how much folding force you have. So how many newtons am I actually holding this with? And how much does it weigh? Remember we do paper clips to hundreds of tons. You know, paper clips probably don't matter. But when you get into the tons, tons matter. What's underneath this hook? So that's where we want to get to.

Winn:
Plus, if you want to simulate the parts, like you were saying earlier, you need that measurement.

Joe:
That's the sticky point right now, the learning point. We're learning and the AI is learning, and the AI is arguably smarter than us and we don't know what we don't know yet. But every time we do learn, we update, we fix. So it's going to be a year, maybe two before we get some of that. And it'll come out and then we'll continuously improve it. And that's the beautiful thing about Magswitch is we're a small global company. We think we're fast to market. If a customer has a problem they can't solve and nobody else can solve, reach out, contact us. If we're dealing in ferrous metal, we can develop brand new magnetic products in sometimes as short as one month. Four weeks. We've done stuff in the automotive . . .

Winn:
It almost compares to mechanical fixturing, the amount of time you'd have to wait for something to be machined.

Joe:
Yeah, exactly, because the magnet's the magnet. We just have to figure out how to arrange it, how to package it, and then actuate it. So we have manually actuated magnets, we have pneumatically actuated magnets, and then we have electrically actuated magnets, not electromagnetism but a magnet that's turned on, think of like by a motor, so we can vary the output.

Winn:
Awesome.

Jimmy Carroll:
Yeah. You mentioned automotive. And obviously automotive has to be a big industry for you guys. But one application that I wanted to ask about in particular is within EV manufacturing. So I imagine you've seen some interest there in some of the specialty tasks like battery cell assemblies. Would that be the case?

Joe:
Yeah, for sure. You know, the cylindrical battery cells have ferrous content in them. And we've done multiple lines for multiple cell manufacturers. Again, the advantage of single-sided gripping: we can pick one of those battery cells up just from the top. The same magnet can pick it up from a row on the sides. So when it comes to loading the modules, so you take six cells and you load it into a module. And then you take that six cells and you load it into a pack. And then at the end of the line you've got an 800-pound battery pack that needs to be picked up and lifted somehow. Well, that's the bigger widget. So we start in the in the cell manufacturing with a little small Magswitch. And then when we get to packs, it's a bigger Magswitch. And then the bundles is a couple of bigger Magswitches. And then at the end we have a battery pack the size of this table that probably weighs 800 pounds. And we pick that up and load it into the AGV or into the car or into wherever it's going to go. So we've been fortunate to work with some really great companies in the EV space and battery manufacturing, as well as vehicle assembly, when it gets into the assembly plants. And then for us, obviously, the construction of the car body, as long as it remains ferrous, will still be in that world of the EV marketplace as well.

Jimmy Carroll:
Interesting. I wanted to ask you about collaborative robots. Obviously, what we've talked about so far there's a lot of large robots and some smaller ones in the bin picking with smaller metal parts. But do you find that integrators and OEMs are using your products with cobots too? And if so, what do those applications typically look like?

Joe:
Yes, we actually have a whole lineup of cobot- or collaborative robot–specific tools. We have end-of-arm tools. We have magnetic bases. So the applications are most commonly machine tending. So loading your CNC, loading your lathe. You have a tray or a bin of parts. The cobot reaches in with a magnet, the Magswitch. We grab the bin, we go down, we load the CNC, we flip around, we take the one that just got finished off. We drop it off on the to-go cart. That's a lot of our applications. I mentioned our magnetic base. There are industrial agricultural companies that need to repair bulldozer blades or excavator scoopers. They all have reinforcement weld ridges on the side and on the bottom because that's a wear surface. With a cobot, a welding system, a generator, and a magnetic base, you can put the magnetic base inside a D9 bulldozer scooper, mount your cobot, fire up your generator, power the welder, and teach it how to weld those reinforcement ribs across the front of a bulldozer blade. And it can do that for the hour or two while the human is doing something else, repairing the GPS tracking system or calibrating the GPS tracking system. So when it's ready to doze, you know everything's all set to go. So deploying cobots in a field environment with magnetic bases is exciting. We use magnetic bases on the sides of ships. So cobots on the side of ships, polishing, sanding, grinding, painting, welding. You know, we have long seams in ships and you can just plop your cobot there.

Winn:
And I think there's a gap of like 20 to 22% predicted for welders, going into 2030. So, again, not a question of displacing anybody at all. It's just about trying to fill those labor gaps that are there, and as you said, put the human where the intelligence is needed: the GPS, the electronics.

Joe:
There you go. And when we talk shipbuilding, submarine building, nobody really wants to be in those cavernous things upside down. If you can just set the robot, and nowadays with the collaborative teach buttons, drag it to where it's got to go and hit the speed and let it do its thing for the next 30 minutes, because it's going to take that long to get through that corner of weld and two passes or whatever it is. I'm not a welding expert, but it's way better.

Winn:
I was just reading last week that the shipyards are super far behind, especially the U.S. Navy has ordered a bunch of ships, but every one of them is years and years behind because they don't have enough machinists.

Joe:
I believe that, and I think you can see it. There were commercials for US Boat, build submarines.com. No plug, no official affiliation.

Winn:
Yeah.

Joe:
Certainly all those industries need help, and when they can't find the people, they're going to be coming to the group here to help them solve the people shortage. the skill shortage. And it's an amazing group of people here, from robots to vision to clamps to magnets to suction cups to AI minds. Some of the greatest AI minds in the world are here this week.

Winn:
It's really cool to see a convergence too of these different technologies. Everything used to be so standalone. But now we've come a long way from vision guided robotics.

Joe:
There's a jingle back home. I'm from Michigan. "You can't do one without the other." We can't deploy randomized bin picking without vision guidance, without super-strong robust robots, without AI language models in the background trying to figure out the path planning and then which is a good pick and which isn't a great pick. So together the cohesiveness is changing the way things are done. And that's kind of a Magswitch tagline you'll see on some of our stuff. But again, it's not just us, it's everybody that's working together. The collaboration within the industry, as you said, is probably greater than I've ever seen it. And as we talked before the podcast started, I've been doing this for 30 years. We used to be silos and standoffish. And if you had a partner, it's like a good old boy. Hey, Bill, I got you. You got me. And now it's just, Hey, let's do this together because we can.

Winn:
Open interfaces and that ability aren't really luxuries anymore, right? Something you just do for large-volume clients. There's just too much demand. Even as a marketer, we can't even keep up. I think we should give a quick shout out to University of Michigan.

Joe:
That's right and the national champions and the Detroit Lions on their way to their second home playoff game in 50 years. So excited for that. It's been a good football season for those of us in the great state of Michigan.

Winn:
Yep. And we got the sports networking event. Is that tonight right.

Joe:
Tonight.

Winn:
So it's going to be a good day here at the A3 Business Forum I think.

Joe:
Absolutely. Jim, anything else on your list? I could go for hours.

Jimmy Carroll:
One thing I will ask you is, I was just thinking, you're talking about the idea of the convergence of technologies and companies partnering together more to solve problems for customers. And I guess one space that you can see a lot of these technologies come together is in the warehouse. Market intelligence company Interact Analysis predicts modest growth for warehouse automation in 2024 but double-digit growth in the years to follow. Obviously, there's a lot of different types of applications within the warehouse, from OCR and barcode reading to dock door scanning and some of the things that you mentioned: bin picking, palletizing, depalletizing. What do you think from your perspective some of the growth drivers are for warehouse automation?

Joe:
Well, I think it just starts with demand. Like 20 years ago, I don't even know if Amazon existed. I think it was a bookstore 20 years ago, and my wife wasn't buying four packages a week. So number one is demand is going to drive. People just don't go to brick and mortar much anymore, which is a shame. I'm from a small town in Michigan, and I love brick and mortar. I love going to see Bob at the corner hardware store. But that's a changing demographic. And I'm an old guy and the young guys just shop online. So demand is number one, right? So there's going to be more and more warehouses. And as more and more warehouses proliferate, as we already talked about, there's a lack of people that want to work. They don't want to do those jobs, for whatever reason. So they'll be automated solutions: AGVs, ARMs, robots like automatic storage and retrieval systems now that have arms and manipulators on them, vision guidance, object recognition. The robot goes up to A-72 inch in this gigantic ARS system, but what if the wrong thing's in there. The recognition system will realize, okay, that's not a bottle of Dawn dish soap. That's a bottle of Tide laundry soap. Wrong one. Somebody's got to fix that. But it's driven by demand, as are most things.

Winn:
The modern world, the whole commerce, the whole supply chain and everything. Even we were in zero unemployment, I'm still not sure we'd have enough to keep up with the world's demand right now.

Joe:
I agree.

Jimmy Carroll:
And it's just easier to use. I was thinking about this. You talk about the teach pendant and how relatively easy some of these products are to use. Companies like yours and all the others here are just making adopting automation easier.

Joe:
Absolutely. It has to become easier because for so long it's been hard. You had subject matter experts that knew model robot A that didn't know model robot B. And now thanks to a lot of software companies that are making operating systems that work across multiple robots, you only have to learn one. Thanks to collaborative robot manufacturers that are literally making it like third-grader-easy to see the robot, point the robot, drag the robot, click the button. It's so much easier than it was before when you used to look at this teach pendant and go, I don't know what RZ plus means. Is it a righty? Is it a left-hand rule? Which one is it? And now nobody has to know any of that. You can just either drag the robot or visualize the robot or click the robot on the screen and move it with your finger and oh my God, there goes the robot moving exactly where you finger went. You know, there's buttons that say "open clamp," "turn magnet on," "turn magnet off," where before you needed an electrician to go land wires. Is this the right output? Oops. Not the right output. It was a challenge. Robot integration 20 years ago was difficult and challenging and arguably expensive.

Winn:
And the reason why only 10 to 20% of companies even came close to using it. Maybe we took a choke point out.

Joe:
And now it's just smarter minds, more creative people understanding the marketplace. Like Joe's Lawnmower Shop wants an easy robot to deploy to load their CNC machine for six hours a day. Yeah, let's work on that. And now it's done.

Winn:
And Joe doesn't have an engineer on staff.

Joe:
Yeah.

Winn:
Or a robotic engineer or a robotic guy.

Joe:
Right. And now with AI coming, that whole thing is probably going to even get easier. I can't wait. There's been more development in this industry probably in the last 10 years than there was in the previous 30.

Winn:
Absolutely.

Joe:
And you know, I'm the old guy, right. But I want to be around for another 30 years just to see what happens. Absolutely.

Joe:
This is the most exciting time in our industry ever, and I'm looking forward to seeing what happens next.

Jimmy Carroll:
Absolutely, Joe, I'm with you. I can sit here and talk more all day. But before we get bombarded with the post-breakfast rush, I suppose we'll wrap it up. But Joe, I want to thank you for your time. I really appreciate it. It's been a pleasure talking to you. If people want to learn more, is it magswitch.com.

Joe:
Magswitch.com. All the information's there. There's a little chat bot you can engage. You can send emails direct to us or all our contact information. We even still have a telephone. You could call on the phone.

Winn:
Someone will actually answer it, and they'll be a human being.

Joe:
At least still to this day, for now. But I appreciate you guys. I appreciate Manufacturing Matters. I appreciate the A3. Let's go see what's going to happen. This is going to be a great time to be in industrial automation.

Jimmy Carroll:
Absolutely, and if anyone has questions, comments, wants to reach out, they can do so at Manufacturing dash Matters.com. And thanks for watching.

Sonix is the world’s most advanced automated transcription, translation, and subtitling platform. Fast, accurate, and affordable.

Automatically convert your mp3 files to text (txt file), Microsoft Word (docx file), and SubRip Subtitle (srt file) in minutes.

Sonix has many features that you’d love including automated subtitles, automatic transcription software, collaboration tools, advanced search, and easily transcribe your Zoom meetings. Try Sonix for free today.

Jimmy Carroll: [00:00:06] Hi everybody. My name is Jimmy Carroll. I’m the vice president of operations at Tech B2B marketing. I’m here today at the Manufacturing Matters podcast, at the A3 Business Forum in Orlando. I have the pleasure of being joined by Joe Cyrek, the president of Magswitch Technology, and my colleague Winn Hardin. Joe, thanks so much for taking the time.

Joe: [00:00:22] Thanks for having me. great event. I guess just quick shout out to the A3, Jeff Bernstein, and his whole team for just being consummate professionals and doing so much for the membership here. Really appreciate the opportunity to be here. And thank you guys for inviting me on this beautiful Tuesday morning here in Florida.

Jimmy Carroll: [00:00:42] Absolutely. Yeah. As long as you’re indoors. Joe, so for those who don’t know, if you could give us maybe a quick primer on what Magswitch does and what you do there, that’d be helpful.

Joe: [00:00:52] Fantastic. Yeah. So I’m the president of the Magswitch Automation Group. So Magswitch has been around since the early 2000s. And we have technology that can turn a permanent magnet on and off. So here in Florida, many of us have the little Mickey Mouse Walt Disney World magnets stuck to our fridge. We have a way to make that not a magnet. Turn it off, turn it on, on a bigger scale. But so we make switchable permanent rare earth magnets that are not electromagnets. Use cases are robotics, automotive, heavy industry, agriculture, railroad, anything that’s ferrous steel. We pick things up and put things down on the end of a robot. We’re not robot integrators. We’re not robot manufacturers. We make the widgets, but they go on the end, and we make widgets that can pick up a paper clip. And we make widgets that can pick up the side of an aircraft carrier. 

Winn: [00:01:47] Good Lord. So I’d like to see that one in action. Yeah.

Joe: [00:01:49] It’s a big widget.

Winn: [00:01:50] I got to check the YouTube channel. Yeah for sure.

Joe: [00:01:53] So that’s us.

Jimmy Carroll: [00:01:54] From an application standpoint, you touched on some of those from a high level, but I want to ask about a few in particular. One is a rise in application-specific systems that come purpose-built for specific applications like bin picking, for example. So you’ve got a robot that comes out and it’s incorporating 3D vision and software, sometimes AI based, and then some sort of robotic end-of-arm tooling. Do you see your products being used in a lot of bin picking or pick-and-place type applications?

Joe: [00:02:24] Uh, 100%. It’s actually been our biggest growth segment over the past 24 months. And one of the unique reasons that a Magswitch is so great at bin picking is because it’s a single-sided picking device. So if you picture a 4 foot by 4 foot by 4 foot deep bin full of randomized metal parts, an overhead vision system finds the part. It tells the robot where to go get it. But all those parts are just in a mess down there. And so traditional methods, somebody would come in with a two-finger gripper and try and grip it, and maybe the finger would hit the part next to it and then move the part it was going to get. So in our vernacular we call that hugging. So if you’re using grippers or clamps, you have to go in and figure out how you’re going to hug the part. With a Magswitch, we’re a single-sided gripping device. We just hover down, go where the vision system tells the robot to go, and then we turn the magnet on and we like to say we kiss it, and then we extract it from from the bin. So a single-sided gripping device has huge benefits in the bin-picking world for chaotic parts.

Joe: [00:03:29] And we’re just one piece of the puzzle, but we think our piece enables a more complete bin pick and enables an easier programatic setup for the robot programmer, for the vision system, and then our tools, some of our tools also have onboard intelligence. So as we go down and we kiss that part, our tool will tell the system that, yeah, hey, you have the right part, that the part is currently all there and present. Sometimes parts have brackets that are added or weld nuts or features. So our tools can tell the difference if we have one part or two parts, if it has the hole in the part that the part is supposed to have, and it has the bracket welded to the part that it’s supposed to have, because we can measure the density of metal through the magnetic field and give that information back to the robotic system. In the event that we did pick a wrong one, the robot can then turn the magnet off, drop it, and we can go find a right one or hold it, put it in a reject bin, and then start over again.

Winn: [00:04:30] So just change the orientation of the product in case it’s got multiple pieces.

Joe: [00:04:34] So in automotive it’s been it’s been a very popular option for Magswitch for the past 24 months. And then we’re seeing things in the steel industry like steel service centers, forging centers, where after the grinding and polishing, it’s in a bin and maybe they need to pick it out of the bin and load it onto a pallet and then shrink-wrap it, and so we can hold that stuff. So that’s a little bigger tool. So with the current generation maybe not wanting to come to work so much, the automated bin picking and line-side loading, where whether it’s line side or if you have a little farm, the company has a farm that loads all these carts, and automated guided robots take it to the line for delivery. It’s going to be a very successful growing field for us and for everybody here at this conference, from the vision guys to the robot integrators. The downside for us, I hate to mention it, but fair is fair. Aluminum we can’t do.

Winn: [00:05:41] You need good conductive materials.

Joe: [00:05:42] If I had the aluminum magnet, we’d be doing this on my boat somewhere in the Gulf of Mexico, but I don’t have that yet. 

Winn: [00:05:49] We got to have something for next year. Joe, can you give any kind of guidance about the specificity, either spatial resolution ability of kissing versus hugging? I mean, obviously you can go small, but it’s just in terms of pick accuracy, maybe repeatability. Is there any kind of general guidance you give folks?

Joe: [00:06:05] We tend not to. So the other cool thing about our magnets — a magnet has a north pole and a south pole, right? Our north pole and our south pole are actually added-on features to our magnet housing. And they can be contoured to any shape so we can match geometry of a truck floor pan. We can match geometry of a washing machine support bracket.

Winn: [00:06:28] Kind of combining a fixture with the magnet itself.

Joe: [00:06:30] You got it 100%. And then we can also add a pin in the middle, a non-ferrous pin between the north pole and the south pole that acts as a guide. And so from a vision system or a bin picking, if we get close enough, we sometimes equate magnets to the old “Close enough is good enough,” like in a hand grenade. It’s horseshoes and hand grenades and Magswitches. You get within a half an inch of your part and you turn the magnet on and like the Star Wars tractor beam, it sucks it right up into your locating pin and into your into your fixture magnet. So huge advantages there that literally can’t be done with traditional clamping.

Winn: [00:07:07] What’s the financial argument you make? You say, okay, we can combine basically gripper with fixturing systems, whereas normally you might have to do a whole separate mechanical systems. Is that  a driver for folks to adopt this technology too or is it just to increase throughput, accuracy.

Joe: [00:07:24] It increases throughputs. So on the bin-picking world, everybody wants to get 99 to 100% bin complete, right? Because if you take a bin away and it’s still got 20 parts in it, that’s no good because now somebody still has to dump that. So the availability to get 99 to 100% of the parts, and bin picking is very important for us and our customer base. and the customers enjoy that. When we come to fixturing and putting Magswitches in fixtures, whether operators are loading parts, it’s actuation time. Our magnets for the most part turn on in 250 milliseconds. So a quarter second on, quarter second off. There’s no kinematics. So if you’re designing the fixture that’s getting loaded, you don’t have clamp open, clamp closed. If it’s an operator loading the fixture and the safe distance to walk through the light screens to allow the safety, there’s no movement, so you can shorten the safe working distance. Shorter light screens, shorter walk path, faster cycle times.

Winn: [00:08:20] Better for retrofits, especially in space-constrained operations, which almost, what 75–80% of your installs probably, if not more.

Joe: [00:08:27] And then the easiest one is if you have a robotic system picking up steel parts and you’re over cycle by two seconds. And you have big clamps, and nothing against the clamp people, and it takes a second to close. Switch those clamps to magnets and you just save three-quarters of a second on the open, three-quarters of a second on a close. There’s the low-hanging fruit of 1.5 seconds in your cycle time and per robot segment or station.

Jimmy Carroll: [00:08:53] Well, it seems like another value there that you guys might be able to add is you probably don’t need vibratory bowls or shake tables as often for small parts, right? You kind of remove that need.

Joe: [00:09:07] It’s a great question, and absolutely. So it’s two basic schools of thought with our single-sided picking. So again we have this bin of parts, small brackets. Vision system tells the robot to go. The robot goes and gets one somewhere in the center. Next vision cycle comes. The vision systems typically want to find something near their origin. So you’re always going into the center. So we’re going down, we’re picking one, we’re picking one, and we’re digging a hole. The other school of thought with Magswitches is you go down, you turn the magnet 100% on, and you just got a barrel of monkeys. Your one part that you wanted is stuck to the magnet, but you have all these other ones, and now you know that to keep that one part, I need to turn my magnet on just to 22%. So you turn the magnet down to 22%, and then the barrel of monkeys is flattened out. So it’s still holding the one you wanted and you go do the thing you wanted, but now your surface for your bin pick has been redistributed and reorganized, so you’re never going to dig a hole in the center. Every time you drop that barrel of monkeys, it flattens out. So instead of, as you pointed out, doing the shaker table or picking the bowl up and swirling it around, you don’t need to do that. Save cycle time, save cost. We just reorganized the whole deck of cards for you. 

Winn: [00:10:22] The fast switching, and I assume this is part of that intelligence. Is that AI based? Is that a deep learning model?

Joe: [00:10:28] It’s not yet. That’s in our product development road map. We have the circuitry and the computing power, if you will, inside there. AI is on our road map. We’re just in the infancy stages of learning what we don’t know. But right now, we can tell you when we pick that part that we have the part, it’s the correct part, that the high-low operator didn’t load the wrong bin of parts and we just picked one that’s a half a mil thicker. So sometimes Model A, Model B, Chevy to Cadillac may have different thicknesses of metal. All our tools can tell those different thicknesses. So we know if this is the Expedition or the Navigator or the Suburban or the Escalade. 

Winn: [00:11:10] And most of those times you’re not even having to use vision, right? You’re just using that through magnetic field sensing?

Joe: [00:11:14] Yeah, it’s none with the vision. It’s all through our tools and the magnetic field sensing. And so we can do that also on the fixture side, if you’re loading fixtures, sometimes parts have to be stacked and there’s no way to tell if that part was loaded behind the current one that’s there because you can’t shoot it with a proximity switch. The vision system can’t see it. But our magnetic field detector can determine it. In the stamping world. One sheet, two sheet, three sheet, double blank detection before you load. Nobody wants to load two pieces of metal into a die and hit the go button. Oops, right? We just need to make sure there’s always one one sheet in there.

Winn: [00:11:48] So two or three years ago you started hearing about people who were deploying AI grippers. So in general, can you give us some thoughts as an expert in that side? Is this a real thing? You said it’s on your road map. You know, neural networks came on the scene 15, 20 years ago, and then AI really came on seven years ago or so. And we’re finally starting to find some really good applications for it in volume. How does that fit into the robotic gripper part?

Joe: [00:12:18] You know, for us, the big conundrum or the hardest sticking point is the fact that every piece of steel is different. Like the metallurgy of . . . 

Winn: [00:12:31] You’re going to get variations, natural variations.

Joe: [00:12:33] It’s crazy how many different steels from the steel companies, although they all look the same on a car or on a Bobcat or on a Caterpillar. Yeah, they’re completely different. So where we’re at right now is trying to train the AI to learn what these different metals look like through the magnet.

Winn: [00:12:52] I honestly thought it would be an easier model, because I’m sitting here thinking, you don’t have 100 different defect classes like a machine vision model, but but actually you do.

Joe: [00:13:00] It’s way worse, you know. Look at these tripods with the lights and cameras here today. If I put my magnet on this one made by Manufacturer A, and it’s going to give me a reading that says, Oh, this is 127, and that one looks the exact same, but it was made by somebody else from a different material. And that one’s 142. Now  our goal is to have the magnet pick up the part, tell you how much folding force you have. So how many newtons am I actually holding this with? And how much does it weigh? Remember we do paper clips to hundreds of tons. You know, paper clips probably don’t matter. But when you get into the tons, tons matter. What’s underneath this hook? So that’s where we want to get to.

Winn: [00:13:44] Plus, if you want to simulate the parts, like you were saying earlier, you need that measurement. 

Joe: [00:13:49] That’s the sticky point right now, the learning point. We’re learning and the AI is learning, and the AI is arguably smarter than us and we don’t know what we don’t know yet. But every time we do learn, we update, we fix. So it’s going to be a year, maybe two before we get some of that. And it’ll come out and then we’ll continuously improve it. And that’s the beautiful thing about Magswitch is we’re a small global company. We think we’re fast to market. If a customer has a problem they can’t solve and nobody else can solve, reach out, contact us. If we’re dealing in ferrous metal, we can develop brand new magnetic products in sometimes as short as one month. Four weeks. We’ve done stuff in the automotive . . . 

Winn: [00:14:36] It almost compares to mechanical fixturing, the amount of time you’d have to wait for something to be machined.

Joe: [00:14:40] Yeah, exactly, because the magnet’s the magnet. We just have to figure out how to arrange it, how to package it, and then actuate it. So we have manually actuated magnets, we have pneumatically actuated magnets, and then we have electrically actuated magnets, not electromagnetism but a magnet that’s turned on, think of like by a motor, so we can vary the output.

Winn: [00:15:02] Awesome.

Jimmy Carroll: [00:15:04] Yeah. You mentioned automotive. And obviously automotive has to be a big industry for you guys. But one application that I wanted to ask about in particular is within EV manufacturing. So I imagine you’ve seen some interest there in some of the specialty tasks like battery cell assemblies. Would that be the case?

Joe: [00:15:23] Yeah, for sure. You know, the cylindrical battery cells have ferrous content in them. And we’ve done multiple lines for multiple cell manufacturers. Again, the advantage of single-sided gripping: we can pick one of those battery cells up just from the top. The same magnet can pick it up from a row on the sides. So when it comes to loading the modules, so you take six cells and you load it into a module. And then you take that six cells and you load it into a pack. And then at the end of the line you’ve got an 800-pound battery pack that needs to be picked up and lifted somehow. Well, that’s the bigger widget. So we start in the in the cell manufacturing with a little small Magswitch. And then when we get to packs, it’s a bigger Magswitch. And then the bundles is a couple of bigger Magswitches. And then at the end we have a battery pack the size of this table that probably weighs 800 pounds. And we pick that up and load it into the AGV or into the car or into wherever it’s going to go. So we’ve been fortunate to work with some really great companies in the EV space and battery manufacturing, as well as vehicle assembly, when it gets into the assembly plants. And then for us, obviously, the construction of the car body, as long as it remains ferrous, will still be in that world of the EV marketplace as well. 

Jimmy Carroll: [00:16:45] Interesting. I wanted to ask you about collaborative robots. Obviously, what we’ve talked about so far there’s a lot of large robots and some smaller ones in the bin picking with smaller metal parts. But do you find that integrators and OEMs are using your products with cobots too? And if so, what do those applications typically look like?

Joe: [00:17:10] Yes, we actually have a whole lineup of cobot- or collaborative robot–specific tools. We have end-of-arm tools. We have magnetic bases. So the applications are most commonly machine tending. So loading your CNC, loading your lathe. You have a tray or a bin of parts. The cobot reaches in with a magnet, the Magswitch. We grab the bin, we go down, we load the CNC, we flip around, we take the one that just got finished off. We drop it off on the to-go cart. That’s a lot of our applications. I mentioned our magnetic base. There are industrial agricultural companies that need to repair bulldozer blades or excavator scoopers. They all have reinforcement weld ridges on the side and on the bottom because that’s a wear surface. With a cobot, a welding system, a generator, and a magnetic base, you can put the magnetic base inside a D9 bulldozer scooper, mount your cobot, fire up your generator, power the welder, and teach it how to weld those reinforcement ribs across the front of a bulldozer blade. And it can do that for the hour or two while the human is doing something else, repairing the GPS tracking system or calibrating the GPS tracking system. So when it’s ready to doze, you know everything’s all set to go. So deploying cobots in a field environment with magnetic bases is exciting. We use magnetic bases on the sides of ships. So cobots on the side of ships, polishing, sanding, grinding, painting, welding. You know, we have long seams in ships and you can just plop your cobot there.

Winn: [00:18:58] And I think there’s a gap of like 20 to 22% predicted for welders, going into 2030. So, again, not a question of displacing anybody at all. It’s just about trying to fill those labor gaps that are there, and as you said, put the human where the intelligence is needed: the GPS, the electronics. 

Joe: [00:19:14] There you go. And when we talk shipbuilding, submarine building, nobody really wants to be in those cavernous things upside down. If you can just set the robot, and nowadays with the collaborative teach buttons, drag it to where it’s got to go and hit the speed and let it do its thing for the next 30 minutes, because it’s going to take that long to get through that corner of weld and two passes or whatever it is. I’m not a welding expert, but it’s way better.

Winn: [00:19:39] I was just reading last week that the shipyards are super far behind, especially the U.S. Navy has ordered a bunch of ships, but every one of them is years and years behind because they don’t have enough machinists.

Joe: [00:19:48] I believe that, and I think you can see it. There were commercials for US Boat, build submarines.com. No plug, no official affiliation.

Winn: [00:19:58] Yeah.

Joe: [00:19:59] Certainly all those industries need help, and when they can’t find the people, they’re going to be coming to the group here to help them solve the people shortage. the skill shortage. And it’s an amazing group of people here, from robots to vision to clamps to magnets to suction cups to AI minds. Some of the greatest AI minds in the world are here this week. 

Winn: [00:20:26] It’s really cool to see a convergence too of these different technologies. Everything used to be so standalone. But now we’ve come a long way from vision guided robotics.

Joe: [00:20:34] There’s a jingle back home. I’m from Michigan. “You can’t do one without the other.” We can’t deploy randomized bin picking without vision guidance, without super-strong robust robots, without AI language models in the background trying to figure out the path planning and then which is a good pick and which isn’t a great pick. So together the cohesiveness is changing the way things are done. And that’s kind of a Magswitch tagline you’ll see on some of our stuff. But again, it’s not just us, it’s everybody that’s working together. The collaboration within the industry, as you said, is probably greater than I’ve ever seen it. And as we talked before the podcast started, I’ve been doing this for 30 years. We used to be silos and standoffish. And if you had a partner, it’s like a good old boy. Hey, Bill, I got you. You got me. And now it’s just, Hey, let’s do this together because we can. 

Winn: [00:21:40] Open interfaces and that ability aren’t really luxuries anymore, right? Something you just do for large-volume clients. There’s just too much demand. Even as a marketer, we can’t even keep up. I think we should give a quick shout out to University of Michigan. 

Joe: [00:21:57] That’s right and the national champions and the Detroit Lions on their way to their second home playoff game in 50 years. So excited for that. It’s been a good football season for those of us in the great state of Michigan.

Winn: [00:22:10] Yep. And we got the sports networking event. Is that tonight right.

Joe: [00:22:12] Tonight.

Winn: [00:22:13] So it’s going to be a good day here at the A3 Business Forum I think.

Joe: [00:22:17] Absolutely. Jim, anything else on your list? I could go for hours.

Jimmy Carroll: [00:22:22] One thing I will ask you is, I was just thinking, you’re talking about the idea of the convergence of technologies and companies partnering together more to solve problems for customers. And I guess one space that you can see a lot of these technologies come together is in the warehouse. Market intelligence company Interact Analysis predicts modest growth for warehouse automation in 2024 but double-digit growth in the years to follow. Obviously, there’s a lot of different types of applications within the warehouse, from OCR and barcode reading to dock door scanning and some of the things that you mentioned: bin picking, palletizing, depalletizing. What do you think from your perspective some of the growth drivers are for warehouse automation?

Joe: [00:23:13] Well, I think it just starts with demand. Like 20 years ago, I don’t even know if Amazon existed. I think it was a bookstore 20 years ago, and my wife wasn’t buying four packages a week. So number one is demand is going to drive. People just don’t go to brick and mortar much anymore, which is a shame. I’m from a small town in Michigan, and I love brick and mortar. I love going to see Bob at the corner hardware store. But that’s a changing demographic. And I’m an old guy and the young guys just shop online. So demand is number one, right? So there’s going to be more and more warehouses. And as more and more warehouses proliferate, as we already talked about, there’s a lack of people that want to work. They don’t want to do those jobs, for whatever reason. So they’ll be automated solutions: AGVs, ARMs, robots like automatic storage and retrieval systems now that have arms and manipulators on them, vision guidance, object recognition. The robot goes up to A-72 inch in this gigantic ARS system, but what if the wrong thing’s in there. The recognition system will realize, okay, that’s not a bottle of Dawn dish soap. That’s a bottle of Tide laundry soap. Wrong one. Somebody’s got to fix that. But it’s driven by demand, as are most things. 

Winn: [00:24:37] The modern world, the whole commerce, the whole supply chain and everything. Even we were in zero unemployment, I’m still not sure we’d have enough to keep up with the world’s demand right now.

Joe: [00:24:50] I agree.

Jimmy Carroll: [00:24:51] And it’s just easier to use. I was thinking about this. You talk about the teach pendant and how relatively easy some of these products are to use. Companies like yours and all the others here are just making adopting automation easier. 

Joe: [00:25:08] Absolutely. It has to become easier because for so long it’s been hard. You had subject matter experts that knew model robot A that didn’t know model robot B. And now thanks to a lot of software companies that are making operating systems that work across multiple robots, you only have to learn one. Thanks to collaborative robot manufacturers that are literally making it like third-grader-easy to see the robot, point the robot, drag the robot, click the button. It’s so much easier than it was before when you used to look at this teach pendant and go, I don’t know what RZ plus means. Is it a righty? Is it a left-hand rule? Which one is it? And now nobody has to know any of that. You can just either drag the robot or visualize the robot or click the robot on the screen and move it with your finger and oh my God, there goes the robot moving exactly where you finger went. You know, there’s buttons that say “open clamp,” “turn magnet on,” “turn magnet off,” where before you needed an electrician to go land wires. Is this the right output? Oops. Not the right output. It was a challenge. Robot integration 20 years ago was difficult and challenging and arguably expensive.

Winn: [00:26:24] And the reason why only 10 to 20% of companies even came close to using it. Maybe we took a choke point out. 

Joe: [00:26:31] And now it’s just smarter minds, more creative people understanding the marketplace. Like Joe’s Lawnmower Shop wants an easy robot to deploy to load their CNC machine for six hours a day. Yeah, let’s work on that. And now it’s done. 

Winn: [00:26:50] And Joe doesn’t have an engineer on staff. 

Joe: [00:26:52] Yeah.

Winn: [00:26:53] Or a robotic engineer or a robotic guy.

Joe: [00:26:54] Right. And now with AI coming, that whole thing is probably going to even get easier. I can’t wait. There’s been more development in this industry probably in the last 10 years than there was in the previous 30.

Winn: [00:27:09] Absolutely.

Joe: [00:27:10] And you know, I’m the old guy, right. But I want to be around for another 30 years just to see what happens. Absolutely.

Joe: [00:27:17] This is the most exciting time in our industry ever, and I’m looking forward to seeing what happens next.

Jimmy Carroll: [00:27:29] Absolutely, Joe, I’m with you. I can sit here and talk more all day. But before we get bombarded with the post-breakfast rush, I suppose we’ll wrap it up. But Joe, I want to thank you for your time. I really appreciate it. It’s been a pleasure talking to you. If people want to learn more, is it magswitch.com. 

Joe: [00:27:46] Magswitch.com. All the information’s there. There’s a little chat bot you can engage. You can send emails direct to us or all our contact information. We even still have a telephone. You could call on the phone. 

Winn: [00:28:01] Someone will actually answer it, and they’ll be a human being.

Joe: [00:28:02] At least still to this day, for now. But I appreciate you guys. I appreciate Manufacturing Matters. I appreciate the A3. Let’s go see what’s going to happen. This is going to be a great time to be in industrial automation.

Jimmy Carroll: [00:28:20] Absolutely, and if anyone has questions, comments, wants to reach out, they can do so at Manufacturing dash Matters.com. And thanks for watching.