Automation in Action: Robotic Pallet Movers are Enhancing Efficiency and Transforming the Modern Warehouse

Hi. My name is, Danielle Bijette. I'm out here in Los Angeles, California. I have been inside of material handling and automation in the warehouse for twelve years now. Started off total warehouse, including conveyor systems and storage media and big integrated systems. And then slowly throughout my career, I've gotten more and more focused into industrial robotics. I also did a, but I tried my hand at making my own start up. It was a software as a service company trying to help food businesses, be better. But I am back, in the world of robots, with, Formic, which is doing robots as a service, and I'm having a really good time. I went to Cal Poly San Luis Obispo, got an engineering degree, and, I've lived out here in Greater Southern California ever since. Awesome. Well, thanks for meeting with me today, Danielle. And I wanted to pick your brain about robotic palletization. We get a lot of questions about it. Mhmm. And I thought, well, what better way to talk about this subject than with an actual expert that works with it day in and day out? So I got a couple of high level topics. I wanna hit you with some questions. Some of them are softballs, maybe some of them aren't. I don't know. Great. We'll just kinda work through it like that. Joe and I have worked together at various companies, in the past. We've actually done pleasure. Yes. We have we have worked together twice, actually, in the past. We let's see. I think you were my project manager for one one something Yep. Long time ago, and then we were colleagues. So we we've known each other for eight, ten years? Yeah. Something like that. Yeah. We've we've done all kinds of wild stuff. All sorts of wild stuff, but we've never lived in the same town. The first big question is, like, what is a robotic palletizer? If you had fifty words or so, how would you describe it? This is how I explain it to my friends who are nontechnical or, like, my mom. When they ask, they go, oh, you're an engineer. That's so great. What do you do? I explained it to them as, you know, in, like, a warehouse or a factory, you have to pick heavy things up and put them on a pallet. So bags, boxes, you name it. Pretty much, it's just a big robotic arm. I usually make this little motion with my hand like the claw, and I go it picks it up off a conveyor after it's fully done, and it puts it on a pallet, and then that pallet goes out into the world. That's that's typically how I explain it to to random people I meet. Okay. And then, like, would you consider, like, some other, like, ancillary palletization or robotic palletization opportunities as, like, pallet to pallet, conveyor to pallet, maybe robot to robot to pallet? Like, are there some more exotic things that still fall into that category? Oh, sure. I mean, you can really sort of is the world is your oyster. Right? You can have multiple lines coming in, and multiple pallet build locations going out. You can depalletize things. So, right, if it's coming in on a pallet, you can take it off the pallet and put it on a conveyor line. You can have upstream automation, you know, pick pack ship, like a pick module upstream coming down to a case. You absolutely can have, like, a robot on a rail that's picking things up from various places and putting them on a pallet. You even can have, robots picking up, a single item and putting it to another outfeed. Just pretty much anything that a big giant arm in the sky, could do, that that's in the realm of robots. Okay. Cool. So so what are the main components? Like, let's kinda talk about how does a palletize what are the main pieces of anatomy of a system, like a robotic palletization system? So first thing is you have to have a way to get it in there. Typically, that's conveyor. I'm sure that I I have done one application where it was basically it came in on sort of like a a a slide, so, you know, a nonpowered conveyor. But, typically, it's going to be a conveyor on the infeed, so you gotta have some way to get it inside of the cell. And then you have to have a robot arm, this this part of things, physically moving. You know, those things tend to be pretty pretty large. They have some some little small ones, but most of them tend to be, you know, in the hundreds of pounds. And then you have to have an end of arm tool, which is kind of like the hand. This is the most custom part of of things. You have a different end of arm tool or different hand for every application. So it's going to look a little different if you're just doing fifty pound bags or if you're doing, you know, two pound little cases. So, you have to have an end of arm tool. Typically, you have to have some way to put things on a pallet. So sometimes that's as simple as we bend a piece of metal just like this and we slide a pallet up into it on the floor, and the robot can absolutely, put things right there on the pallet. If we get a little more advanced, we can start doing some special handling. You can build a pallet on top of a conveyor so it can come in and out automatically a little bit faster, a little bit more equipment. You can also get fancy and have a pallet dispenser. So you load up a big stack of pallets just like a PEZ dispenser. And then when a pallet is needed, it just spits out a pallet and goes to where it needs to go. You can also get fancy and have a bottom sheet or slip sheet dispenser. You can have that be put just like a PEZ machine right on that pallet, or usually, the robot can actually go down to a magazine, have a little vacuum, like, nozzle, and then pick it up and put it on. So you can do some some spicy things. You also need to have some safety. So sometimes we put a fence around it. Sometimes we do a fence and some light curtains. And with the fancy new, collaborative robots, sometimes you don't need a fence at all. So, that that is, sort of at bare minimum. There's a control panel. There's some electricity stuff. But pretty much get it in, get it out, have a robot arm, and have a way to control it. Okay. Cool. And then part of the anatomy that may this may be a little bit more invisible, but I think for folks like us, it's very much important is, like, how does it know where to go and how important is presentation of the case or of the palette, right? Because a couple of things you mentioned are actually, like, indexing mechanisms. Sure. What what does all that look like? So most robots, are blind. There are robots that have vision and cameras on them. They're they're typically a little bit more expensive, and they have a reason for that. But most palletizers are blind, meaning they don't have a camera, they can't see, but what they do know is x, y, and z coordinates. So when that case or that bag comes into that infeed conveyor, typically what's going to happen is it's going to be justified to the same exact point. Right? So it's always gonna go whether it's mechanically a we, you know, run the box up to a little t stop kinda like this. Stops. Yep. Just boop. And it just goes right there, then robot knows that's that's where it's supposed to be. There are a little bit more sophisticated ways, we can use lasers to justify things. It kinda depends on what the end of arm tool is doing, whether it's like a top down vacuum or a clamp or a little combo of both. But we have to basically take the thing we're picking up to roughly the same spot in in the universe, the x y z. Right? So because the robot's blind. So it's just gonna blindly pick that thing up at the same place and then also has to place it. So same thing, we wanna make sure that that pallet is in the same spot in the x y z world, and the robot has to also keep track of where it is on the pallet pattern. So it has to have a memory. It needs to have a recipe. It needs to have a layout of, okay. We have, you know, twenty boxes. Right? Four this way, five that way. Where am I? Am I on layer one, two, three, four? So that part, the robot, again, is blind. So it it has to keep track of how it's stacking everything. So, like you were saying, Joe, getting that in, and justifying it correctly in in your world, does the sort of technology, is it also blind or, like, do you have to justify things in the same way or kinda how does that work? Yeah. Unlike conventional conveyor systems, presentation is kind of everything, right? Like you need to make sure that when I enter a transfer or a divert that I'm in a position that I can actually like move the product in a positive method or like way from there. So it is very important. Now there's some logistics circumstances where it just kind of you can't do that. So I think you and I have also done a couple of bulk movement projects together. And it's just like this big fat conveyor, and it's just got random stuff in it. Right? Have you ever worked on any, like, like working from or I think you mentioned you had a slide, and that's a kind of an uncontrolled, pickup point. Like, talk about that a little bit. It was the slide. It it was a semi controlled pickup point. So it would come down, and it actually would hit, like, a a a stop. We didn't have control over it left and right, but what we did is we used, a clamping tool. So, instead of coming down at the top. Right? So if you imagine, like, a big you know those claws overhead, for, like, the the the the game you play at the arcade. Yeah. You pick up the ball and you get a prize and you always lose. Yeah. So think of it kind of like that, but, you know, with better odds, because we're gonna pick it up every time. So because we couldn't control, the where the box is left and right, because it was a slide. Right? Yeah. Some end stops, but it it would move, you know, three, four inches. We had an end of arm tool that was kind of a big claw. So it had lasers, looking, down inside of the the, the slide. And when the lasers were cleared, right, so we knew it was in, you know, roughly the center, The clamp tool would come down, and it was bigger than the box and would kinda just go, boop, and pick it up. So we we have the ability to get a little bit, I don't wanna say sloppy, but less precise. So You could you could compensate for that that that y value that wasn't precise every time with, like, a larger, maybe more expensive, like, tool? It was. It was a more complicated, more expensive tool. The good news in that was we already had to kind of use that tool because the boxes were fifty, sixty pounds, to start hitting the limit of, what sort of off the shelf vacuum tools can do when you start getting that that, that heavy. We like to start supporting it in a mechanical way instead of just relying on that vacuum for those it's just when you pick a trash bag up and the bottom folds out. Right? You can do the same thing with the box. Exactly. As that. Okay. Makes sense. Yeah. That's precisely it. Once you start getting really heavy, like, you know, we we kinda want something underneath there or something, you know, mechanically grabbing it. Not to say that vacuum, you know, can't be used, but, it it depends it it puts a lot more onus on the box, to be good. So, anyways Yeah. We already had to use that clamp tool, so we're like, oh, this is fine. We have, you know, we have three inches of play on each side. It's no worries. Alright. Cool. So, like, is it is it generally that, like, the more freedom you allow in that presentation, like, the higher the cost is gonna be for the solution? There is a pretty linear correlation. Sort of the the more variation and the more sort of complexity of how the item is presented in general is going to have some, some cost implications. So I always try to make it as simple as possible. How can we put a bent piece of metal at a precise position? There's, you know if you're breaking, you know, steel, we've got big issues, but it doesn't require any maintenance. It's very, very affordable. It's very, very repeatable. So when I can get away with it, I try to just make it good old fashioned. You go up to a little end stop like this, and your box, you know, is justified. And, you know, you you got a powder coat, a piece of steel, and you're you're ready to go. Alright. Cool. Cool. Yeah. Okay. Awesome. Let's also talk a little bit about, like so right now, we're talking about, like, one box. I take a box, I move it. I take another box, I move it. And I think this applies a little bit more to depalletization. But do you ever handle multiple units? And then, like, what's the thought process there? Yeah. So, there's some pros and cons. When you start handling multiple units, obviously, you can go faster. If you're just like, you know, at home, if you're picking up two things at once, it's you know, you're going twice as fast or, you know, three times as fast. The the cool news is you get a lot more rate for almost free. Right? So you're moving the robot the same number of times, but you're now picking up two boxes, three boxes, four boxes. It's almost free. So what's what's the cost? It's almost great. So, if you're picking up two or maybe that pushes you to the next, you know, bigger payload. So, you know, the actual robot arm is larger. It's more expensive. Could be that. You also now have two sort of zones. So you have a larger end of arm tool, more money, more robust. You know? It it does take a little bit more complexity, and I say almost free because depending on how the pallet pattern is, sometimes they're really nice and easy where it's just, you know, four boxes like this and then one more boxes like this. But sometimes, depending on if it's, like, interlocked or if it's a complicated pattern, sometimes it's one box like this and then another box ninety degrees this way and another box ninety degrees that way. So even if I'm picking up two at a time, depending on where I am in the palette, that means I may have to take those two boxes. Oh, I have two phones here today. Look at that. I may have to take these two boxes. Right? Drop one of them this way, pick it back up, and drop the other one that way. So Okay. Yeah, so you so you get that. It really depends on the palette pattern and how fast you have to go. It also does require a little bit more complexity of the, programming because, again, yeah, like, let's say, you know, you have one this way, and then let's say the next box is actually the next layer. So it does require the, the robot cell to have the ability to drop one at a time sometimes. Some are very straightforward, just two by two, and that's fine, but little bit of variation. It's a pretty cool way to get more rate out of something without a whole lot more equipment. Okay. Very cool. Very cool. Good. And then you talked a little bit about, like, putting the robots in a cell and how well safety around it or a cobot. Right? So a robot that an operator can safely, like, be around. And and sometimes that would be desirable if I wanna go and change out a pallet position because that one pallet's done being built. I wanna put a new one in there versus having to put in a lot of, like, pretty intense pallet conveyor. Because sometimes sometimes it's it's oh, there's a lot. So talk to me about, like, cobots and my what the trade offs are with a traditional robot, and and why I would or wouldn't wanna, you know, look at that for palletizing. Yeah. So, Joe, if you put yourself in the shoes of a customer. Right? If I was to say, hey, Joe, we're gonna we're gonna make you a robot sell. I know that you've been in this plant in New Jersey for, like, eighty years. Right? This is your great grandpa's pickle company. And you're like, yeah. We have you know, we are interested in putting a robot in, but we are really tight on space. And you're like, okay. Cool. So what we can look to do is try to put, a two pallet build location, you know, something like fourteen feet by fifteen feet, something like that. And, you might say like, oh, okay. We we got enough space. There's a column there. Like, you know, we we can make it work. Everything's cool. And then I'm gonna be like, Joe, we have to fence off pretty much the entire section except for where you're going in and out of the cell, where those pallets are coming in and out. And you might be like, oh, Danielle, that's not that's not gonna work. We have an operator that, you know, he he goes from the filler up to, the palletizer, and there's no way for him to, you know, get all the way around there. You're gonna want a monument. Right? Yes. It creates a huge monument, and it's visually very impactful. I always try to preface people to say, like, I'm going to be putting six foot tall fencing inside of your facility where there was nothing before. Your brain is going to have this moment. Every install, there's always this moment where the fence goes up and people are like, oh my gosh. It's so much bigger than I thought. I'm like, actually, it's really not. But if you can imagine same size. Yeah. It's the same size I told you. Imagine. Like, if you can imagine you're in your living room and then all of a sudden, like, you get one of those privacy screens and you put it up in the corner, you're gonna be like, wow. My living room looks so much smaller. And you're like, it's a five by five area, but, like, yes. Visually, it's very difficult. So you have this game you have to play where, traditional automation, we do the fencing and we do the light curtains. Right? Sometimes some area scanners. It's very big. It's visually very big. It can have these big monument style flow disruptions, but they tend to be faster. So a traditional noncollaborative robot cell, they the actual cell, the arm itself can move faster. The reason is, there's a safety equation. It's a multivariable equation. It's like a OSHA rule. It's also from, like, the international I have to look it up, the the name. There's a multivariable equation that has a bunch of components. One of them is stoppage time. Right? So, like, how fast the robot can react. One of them, multivariable equations, one of the variables is how fast the robot is going. Another one is, like, distance, and then there's a couple of other, ways to make it up. So, if your robot is moving quickly, right, it it's just like a freight train. Right? If you have a a big heavy thing moving quickly, it's going to stop slower than if you had that same thing moving Right? Yeah. If you have that same thing moving slower. So our traditional automation, our noncollaborative robots, we we throw a physical barrier, a fence around them. We throw some intelligent barriers like curtains, you know, area scanners so that that robot can move as quickly as it it can knowing we have these barriers to entry. The collaborative robots, they they can have smaller footprints, but the the trade off is since we're not physically putting a barrier around them, the actual arms can move slower. So they simply cannot move in the x, y, z, like, dimensions. Like, their speed is governed by that equation. Right? So let's say it's same exact arm. Right? No difference in mass. The the cobots, do tend to have slightly faster stopping times, but because I have to stay underneath, an allowable sort of number, I cannot physically move that robot arm as fast. So we play this game where if you need to go very fast, I always try to recommend a noncollaborative cell because there are less sort of restrictions on the speed of the actual arm. If if your rates right? Let's say, upstream, you're like, Danielle, it doesn't matter. My pickle guys can only make pixel pickles at six boxes a minute maximum. There is no way that the briner and the cucumber chopper I'm just, like, fully making this up. The black box pickle machine. Yeah. Right. There's no way that we can give you pickles faster than about six boxes a minute. It's impossible with the equipment we have. We're not gonna upgrade for another, you know, twenty years. I would say, what an interesting moment. It sounds like it might be okay to go to these slightly slower, in general, collaborative robots knowing that it will be visually smaller, technically smaller, and we won't have to have the operators be, so sort of removed from interacting with the cell. So it's, what's that thing? It's like cheap, fast, and good. For robot cells, it's like cheap, fast, and, you know, and good, really. It or, like, how close you can get to it. You have to kinda play around and see, do we go down the traditional path? Do we go down the collaborative path? They are getting better. Five ish years ago, ten ish years ago, maybe seven at this point, collaborative robots are really, like, coming into the market, and they were so slow, like, painfully slow. The technology has got yeah. I was like, there's they were so slow. The technology has gotten a lot better where they can stop faster so they can actually move a little faster. But you're you're constantly pushing up against that, you know, OSHA and, robotics, I can't remember what the name, association, formula. So Yeah. Just like everything, it's always a trade off. Right? And, also, there there is one more factor. Sometimes so whoever's doing the marketing for collaborative robots is crushing it. They are whoever, whatever group of people are doing that, good job. Many small to medium businesses, will see a collaborative robot at a trade show on sixty Minutes, on some YouTube video, in a, you know, trade publication, wherever people are getting LinkedIn, wherever people are getting their media. Sometimes in family owned businesses or when you have, like, decision makers that are, you know, not necessarily technical or things like that, they will get this idea in their head that a collaborative robot is the only thing that they ever want. And so you do have to wrestle with that because the marketing conceptually is very, very good. You're like, yeah. I can have this robot right next to this guy. It won't kill him in theory. Put it on my floor. And then you're like, well, if we put a chainsaw in the hand of this robot, he technically could kill this guy. You know? So there's a little bit of of of nuance, but, again, whoever's doing the marketing on cobots, great job. Killing it. Yeah. Okay. You've have you have you ever encountered that, like, in your world where somebody will have a very big bias for a technology even though you're like, this is a terrible idea? But yeah. You know, sometimes we see that with, like, AMRs and AGVs and stuff like that. Like, if you if you bound them, then you can have a machine that goes four meters a second, let's say. But but in unconstrained fields, you you have to dial them back to, like, one meter. Right? So, like, it's a huge differential in, like, throughput that you can expect out of the same, like, capital investment. And so, like like, we definitely do the dance all the time. Like, how how fast do we wanna go? How much do you wanna spend? And how important is that to not have a monument built? A lot of times it turns into a huge track or monument. Right? Yeah. So I yeah. I think it's a it's it's well taken point. So what like, for, for a robotic arm, what's the what's the cost difference for a collaborative robot versus a traditional one, and and what's the rate difference that I might see? In general, depending on how you configure them, they are getting within five, ten percent of each other. So they are comparable. Yes. Cost wise. Okay. Well The reason being is if you have a collaborative robot, typically, the arm itself will be more expensive, but now you don't have to buy fencing, and you don't have to buy light curtains. Oh, okay. They're they're getting close. It it it's there even are some collaborative robot cells that are cheaper than traditional automation when you look at the total package. Depends on, again, weight and rate, but for for sake of conversation, you know, you could consider them the same for, you know, the same. They're they're they're pretty good now, which is which is exciting. Because when they first came out, they were definitely more expensive, but they've gotten a lot better, and they're pretty comparable. And and then rate wise, like, how much of a hit, like, would you expect to take? There's, like, like, relative terms. Sure. I've been working I was just working on a, like, a box palletizer recently with a cobot. Their rules of thumb, were, like, anything less than about thirty five pounds and up to about seven ish picks a minute, seven, eight, depending. That includes a slip sheet. So if you have a slip sheet, that's a pick. And that's a single, you know, single box pickup. So, you know, eight, seven, and about thirty five pounds for a collaborative robot. For the same sort of setup, one in, two out, single box, whatever, you with the end of arm tool, you know, I would say rule of thumb forty five, fifty pounds so you have a little bit more room there. And, you know, something like fifteen to eighteen a minute depending on pallet patterns. So we're gonna, like, twice the speed and and more weight. Yes. So, again, it kinda depends, but the rules of thumb are you can totally do eighteen, twenty cases a minute, with a traditional, robot cell, like, off off, you know, kinda off the shelf. So you you do take a a rate hit of about half. But, again, if your pickler only goes so go so fast, there's no there's no point going any faster. It doesn't it doesn't matter. So it it depends on the application, depends on how fast, you you know, the upstream your pickle machine goes. I would I would lead you down whichever path sort of fit you best knowing that they are almost equivalent in price. But you you do take a pretty big hit on on the rate. Okay. So before we talk about future stuff, I got one other, wild kind of question for you. So so we're you you kind of specialize in robotic palletizers and depalletizers and that there's largely is like there is a an arm has several degrees of freedom, then it reaches out and grabs stuff like a like an arm does. But there's this more traditional, like a like a older technology, right? These unit load formers, for lack of a better word, where you have a conveyor and it slaps the boxes around into a pattern and then just slaps it on top of the pallet and then and it just makes another layer, repeats it. What's, when when should I consider one versus the other or have robots come so far that, like, that's just a thing in the past now? Like, what's the what's the conversation there? Sure. The the gantry style or the lair robots, I mean, those have been around since, I think, like, the seventies. Forever. They came out back then. Yeah. Older than me. Yeah. They are very robust. They can be very affordable. They can also fit in really tight spaces. I think like a layer palletizer or a top down palletizer or like a gantry, sometimes we call them, those non robotic palletizers, I think they work very, very, very well. If you if if your actual, like, layers that you're forming are very simple, meaning, if you have all these boxes just lined up, you know, like, one just in a row, there's not interlocking patterns. There's just, like, good old fashioned column stack. Right? We're gonna put four boxes this way and five this way. If we have something like that where the the actual pallet patterns are straightforward, if you have, boxes that are very repeatable, meaning, like, you only have two pallet patterns. Let me give you example. There is a company that makes, like, eighty ish percent of the mozzarella cheese for, pizza shops, like Papa John's, Domino's, whatever. They just make mozzarella cheese like it's going out of style. I love the pizza. Yes. They're great. They have two. They have two size cases. The end. They've had it for fifty years. They have the big case and the low case. The end. It's column stacked. It comes very, very, very quickly. I think they were in the thirty, forty minute. Right? They're just making mozzarella cheese like it's going out of style. Yeah. And they have very straightforward just, they they only run one product per day. They don't do a whole lot of change over overs. They I quoted a a robot palletizer and also a traditional kind of gantry. The gantry actually was better for them. It was less it was faster. It was smaller, and it was cheaper. And for them, having those degrees of freedom was actually a a detriment. So they work when you have sort of a simplicity in how many products you're doing, and the pallet patterns. They they work really well. I don't shy away from sort of exploring that option when I see just a textbook, very straightforward type of thing. Now robots have gotten, more common, meaning there are less providers of those gantry trains. They tend to be legacy providers, meaning they tend to have a little bit less in programming, remote monitoring, making changes on the fly, not all of them. I'm sure there are people out there that are integrators that are really, really good. But because this technology is from the seventies, eighties, and nineties, the software tends to be a bit more dated than a robot. There's some challenges there, especially when you're trying to like, we do robots as a service. When we're trying to make sure that we have high level, crash detection, monitoring of different parts, sometimes those gantry components simply are less tech enabled, so we can't monitor them as easily. The programming sometimes runs into challenges of making changes because it is, like, a CSV file that we zip into it. I know that sounds wild, but, like, that's the reality. Eighties. You know? Yeah. That's the programming sometimes. Mhmm. There tends to be less human beings out there that know these programs, and it tends to be less visually, appealing on the HMI. So there's a little bit of that, but, I mean, it's outrageously an option, that still exists. It is it's dwindling because robots are so fast and cheap now, but it's it's a great question. If you find the right application, it's still a home run. Yeah. It's still a home run. It's still a home run. I'm sure you have, some sort of legacy, sort of, like, conveyor technology that you use that, you know, has been around since the seventies and eighties that sometimes is the right choice. Yeah. Yeah. Yeah. Absolutely. There's all I mean, we still use shoe sorters. Right? There we go. They're not new. They're fast. Right? Not new. Yeah. Yeah. They they they push a box real good, you know, when when the the rate and the time is is is fine. So Absolutely. Okay. So let's let's, shift gears a little bit and talk about some future stuff. So, like, try to predict the future a little bit, And then let's try to hit yeah. Yeah. Beat your crystal ball. Let's try to hit some stuff like, mixed pallet build. So because this is this is, like, I think, on the hard end of, you know, palletizing. So what does that look like? And then, like, where are you where are you guys slotting here? Where do you see, like, AI helping out and, like, the super advanced vision stuff, really, I think is probably what I'm asking there. Yeah. So, like, like, talk to me about these kind of topics. Alright. Sort of I was explaining this to someone the other day. So with non text so, like, not vision enabled, not AI enabled palletizing, You need to tell the robot this is what we're doing, and we have to do that. Right? So, at your pickle company, you have to tell the robot we are doing the small pickle boxes And this recipe pattern. In this recipe that is half it's set. It is set in stone. Yeah. And that's cool and fine if you make small pickles and big pickles. Right? It's no big deal. Absolutely. Kind of where we're sort of sitting on up in in the robot world is being able to handle variation. So, one of the easiest things you can feel is trailer unloading. So knock, knock, knock. Mister trucker is here. He's got a fluid loaded trailer. Right? So box is stacked. It locks the configuration. Hot topic right now. Yeah. Everybody's trying to do this. Right. Knock, knock, knock. Your container is here from China. It is fluid loaded. You have, you know, twenty seven different box sizes in an infinite configuration because somebody, some humans, somewhere else have put as many boxes as they possibly can. So I'm Yeah. Yeah. You're you're here, and you're like, okay. Might as well just get, like, seven dudes, and we are going to unload this trailer by one by one. Yep. Take a box off, put it on a conveyor behind me, and that conveyor goes and does stuff. So the control kind of sucks. Right? It's a lot of torsion. A lot it's hot. It's terrible. It's not fun. Yeah. Everyone wants to do it. It it's time consuming too. Right? It takes a lot of people. It's mind numbingly boring. It's, like, stunningly boring. Right? So let's throw a robot at it. Okay. Let's try to do this. So now imagine yourself as a robot. You are so first, you're blind, so we gotta make you see. So we have to throw some cameras on you. Right? Now, we have a camera on you, and you need to be able to identify the different boxes in their LEGO configuration. That is a lot of processing. Do we use edge detection? How are we using this camera to make sure that we know this is a box that's on the top right and it's twenty inches by seventeen inches. This is where AI comes into play a lot because, I mean, in industrial applications, AI is just how can we process large amounts of data quickly and repeatably? How can we train something to, like, learn faster? Right? We how can we take bajillion images of boxes and be able to parse down, okay, that's seventeen by twenty inches that's sitting there. So once we enable the robot to see the LEGO configuration, right, using vision and AI, cameras are getting real good and cheap. I mean, we've got in insane cameras on the iPhone. Like Yeah. You know, remember our, like, three megapixel cameras from, like, the two thousands where we're like, I'm gonna upload these on Shutterfly, and it was, like, blurry and terrible. Right? You gotta take the the the card out, put it in your computer and upload it to Myspace, hundred percent. We have we have lenses. We have we have image processing that can go very fast. We throw a little AI on it. Right? So we're we're teaching we're taking millions of photos of different angles of inside of a a trailer that's being unloaded. Once we accomplish that, then now we have to say, okay. I see the LEGO pieces. I see the boxes. Now I have to tell my robot arm, please go here. Use your hands, right, your end of arm tool, and you have to grab that very precisely without knocking everything else and not damaging this box and taking it back to a conveyor. So I think that coordinating that conversation of what me and you as humans do in a moment, right, our brains It's trivial now for people. You you could take a, you know, a six year old and say, hey. Like, take these boxes, you know, from Christmas down and put them, you know, somewhere else. It actually takes quite a lot of computing up up here. How do we trade a machine to do that? So the trailer unloading problem is is huge because, like, into your systems, that's probably a huge bottleneck. Right? Yeah. Well, yeah. So we we we have the all these traditional conveyor unload. We put a flex conveyor or we put, like, a Cal JN unit in there and and RV stuff it. Right? We try to make it as comfortable for the person as possible. But, really, what we're doing is we're making it a very boring hard job comfortable. We're it's it's it's we haven't cracked the code yet. And I think I think that we're very close as an industry. There are there are, quite a few robotic outfits that are tackling this problem of vision, AI, and how do we translate that into repeatable robot arm action. Mhmm. There's probably, like, four or five sort of, at least in the United States, that are doing proof of concepts, like, have actual machines out there. I think Picklebot is actually the name of them. Yeah. Pickle Pickle's one of them. I think, like, Dexterity is also one of them. And Boston have one as well that they're that they're working on? Boston has one that they're working on. There's a couple more inside of, like, the larger system integrator space. Everyone's trying to chop at this. And with the rise of cheap vision and I wanna say, like, cheap processing power, but, like, cheap processing power and trainable algorithms, so that's AI enabled stuff. That is where you're starting to sort of get over that hump of how do I take the box how do I identify the boxes and take them off very quickly? Some people have tried to do it mechanically. Like, there's some old school patents out there where you have the the container, and then they're like, you just tip it. And then, like, we control the fall and singulate them out. Like, that's Yeah. A controlled tumbling. Yeah. Yeah. So there's some singulators actually on the market that work with sort of the same principle. And it's always a conversation, like, every time we do a whole solution is this well, let's let's, automate the trailer in mode because that job obviously looks super easy. But again, for all the reasons you've just said, like, this is the the last tough net to crack in most modern distribution centers. Yeah. And so AI is tackling that that random, you know, depalletizing trailer unloading. Depalletizing is a sort of easier version of trailer unloading. Trailer unloading is they're they're similar. Right? So we're looking at something random. We don't know. We have to identify it, and then we have to pick it up. Depalletizing is a little easier because we only have to do it a pallet at the time at a at a time. Trailer unloading is more complex because we have to do a trailer, and there's, like, a whole level of depth that is is much more difficult. So Yeah. I think depalletizing is further along than trailer unloading, but they are they are sisters. Right? They're they're they're similar. Once you can kinda get one but, yeah, it's it's really that question of vision and AI, how fast and how quickly can it do it repeatedly. So, yeah, that's kind of what's happening right now. We bought pallet bills, a mixed pallet bills. Like, this is the this is the, like, this is the other end of the of the most building. Right? So I my I've sorted all this order and I've got, like, this random assortment of boxes. They're not all the same size. And actually, humans even still struggle with making palettes like a good palette. You'll hear They do. Supervisor say, oh, this is a bad palette. What's bad about it? Like, it's it's just a qualitative problem. Right? You have, we'll call it, like, rainbow pallet or or mixed pallet building. It's it's of a similar vein as depalletizing. You have a couple of advantages that make it sometimes a little easier. One, you tend to have more time to process. Meaning, if we can, like, Cubiscan, which is like a proven technology. Right? If if something's coming at me singulated, I I actually have the luxury of knowing, like, it's, you know, ten by ten by ten, and it's come and go on. Two pounds. Right. And it's thirty two pounds. We can use, you know, a checkweigher that we've had around for, again, like, the seventies. Yeah. Dimway scan technology is, you know, locked in. It And that can transmit very quickly. So then at least we can sort of, like, tell the robot, hey. You don't have to do this with vision. We we're gonna we're gonna tell you. We're going to give you the answer. It's kind of like taking a test and, like, somebody's like, I'm gonna give you, you know, I'm gonna I'm gonna give you part of the answer. It's c. Yeah. It's c. Right? You you have a hit. Right? You already have a hit. So we know if it's ten by ten by ten, which is terrible blocks of pallet ties, but whatever. How do we get on the line? Right? And it's thirty two pounds. Okay. Great. So now that sort of the the the algorithm, the AI, the computing, the intelligence that I'm doing is not the seeing part, but the putting part. Right? So it's like knowing that I have these things coming on deck. Do I know three of them? Do I know two of them? Do I know one of them? We have to figure like, now we have to figure out how do I build, right, the the pallet. So that's that's you you you get the answer on what's coming to you, but you still have the conundrum that a human struggles with is, how do I put these, how do I put these boxes on this palette not knowing what's gonna come next? It's the Tetris problem. Yeah. You're playing Tetris. Yeah. You only know one weird shape ahead. Yeah. Right? So you think, like, oh, I'm not gonna wrong. Yeah. This long skinny one here, and then you're like, oh, well, it should I got another I got another t shaped one. Great. Yeah. Great. So Oh, and another t. Yeah. Yeah. So it's it's the Tetris problem. So, that part, people are are decently good at because you have that extra advantage of knowing what's coming. Yeah. So that's good. Right. Sick. Yeah. That that's my yeah. That's my kind of my opinion. What do you like, what what have you I know we're at that time, like but saying it back to you. From your end, what these are all coming from the robot perspective. What's the most pressing issue when it comes to, you know, total warehouse solutions, you know, your your automation? It's really that, so so getting it off of the truck and getting it, built into a pallet or getting it onto the truck, like, those are the areas where there's not, like, solid solutions today. So everything else in the building, I can look at an array of technology, and I can evaluate which one's the right investment. Right? And which one's gonna be the best fit for this operation and all their data. Right? Like, because oftentimes, I'll I'll have a customer look at their inventory snapshots and look at their order mix. I'll look at their item masters and, like, understand all of the stuff. And and everything between between the doors, it's like there's a there's a plethora of solutions. Right? I can I can go robotic here? I can go mechatronic here. I can just use some traditional conveyor, or maybe it's it's, like, you know, powered industrial trucks. Right? Like, there there's a way to figure all that out. But once it gets to a door, like like, everybody just kinda doing this. Right? And then you'll see, like, the least amount of automation, like, in the entire you go walk through a building that's got millions of dollars on automation. You get to the door, and you're like, okay. Well, there's there's some guys throwing boxes in there. And and that's kinda that's that's probably the biggest, like, conversation we end up having when we go through solutions. Like, I think once that that dam breaks, and it sounds like it's very, very close. Right? Like, all the all the stars are aligning, like, then it's gonna be it's gonna be one. We'll we'll start seeing lights out buildings, I think, once that happens. Yeah. I mean, I think there's always gonna be a couple of humans around, but yeah. It's it's so interesting that the ins and the outs, which are so so important for the performance of the system, are are, you know, in the they're in the dark ages. They're they're still just like dudes throwing boxes.