Streamlined Operations & Enhanced Efficiency: How Tote Pallet Building Systems are Revolutionizing Robotic Palletization in Modern Warehouses

Hi. I'm Joe McGrath. I'm a solutions design lead with High-tech Intralogistics. And I wanted to take a little time today to talk through, some interesting combinations and arrangements of pallet building systems. Not focusing just on the specific robots that are gonna actually be manipulating and moving around items, but but what are all the systems that feed that? And how do you make a a full arrangement of a system that meets the needs of the operation? Okay. So let's talk about tote pallet building concept. So we're gonna focus today on totes. It's a pretty simple, repeatable type of item. Sometimes when we get into cases, we were talking about lots of different variation, different package types, sizes, shapes, weights. So to keep it simple and to focus on the system level, flows today, let's just stick with totes. The automatic building of pallet loads is a pretty important consideration for a lot of operations. It can it can improve many, many areas, of your operation. And think about things like general capacity for the whole building. So maybe you only have ten people that can work your dock, but you need more capacity. And you can build in additional capacity through robotic palletization, or maybe even a more traditional unit load form or something like that, to really amplify what that operation can do. Additionally, like, it can improve the utilization of your labor. So instead of having somebody waiting at the end of a line for a box to come down so they can stack it or tote to accumulate on the line before they walk over and and deal with it, you can have the robot do the waiting. And the robots can have utilization issues as well, but it's not gonna be a labor utilization. You're not gonna be paying for it by the hour percent. So it can really make you can really make sure that your employees are doing the most they can do, and it's gonna be a little bit more satisfying and a little bit safer, operation at the end of the day too. So a lot of these totes and really when talking about palletizing things, these things can be heavy, and they can be, difficult to get onto the pallet. Sometimes you have to step onto a pallet and lean over holding, you know, upwards of fifty pounds. Sometimes the things are heavier than fifty pounds, that you're moving around, and that's actually very difficult to do. And there's a lot of, reaching and twisting and bending that it ends up happening in a pallet build operations, when they're manual. And, you know, a robotic version of this can really clear up a lot of that that issue. And the other thing is when you automate a process, oftentimes, if it's done correctly, you can reduce the number of human errors that, tend to occur. You know, a person's busy or a person misses a digit or something like that, or forgets a scan or they thought it scanned and the RF didn't beep. So you can reduce or, eliminate sometimes, these errors that happen in pallet building process. Maybe you shipped a tote, but you didn't actually manifest that tote, and that that could be a big problem. So there's a way to get around that as well. So here we're gonna discuss, you know, like, a a uniform pallet of toads. We're gonna be building it. We're also gonna be focusing on something that's really a single item. Right? So these aren't, like, different sized totes, and we're gonna be focusing on solutions that are fully automated. And there now there are a range of manual pallet building operations, that I think a lot of folks are familiar with today. Pick up the the tote, you stack on the pallet. There's also some semi automated solutions. We're not gonna dig into them today, but they're also pretty interesting from a a capital expenditure and and return on investment perspective. But today, we're gonna focus on fully automatic systems. So the first kind of solutions that we wanted to look at was a simple palletizer. So this could be something like a single, six axis robotic arm that grabs a tote off of a feed line and then stacks it onto a target pallet. Now the simple systems, are gonna have the more simple robots are gonna have minimal reach. So you're only gonna be talking about one or two target positions really. When you start adding more target positions, the reach of the robot goes up, and it becomes a little bit more complicated, a little bit bigger, a little bit more consideration in the the speeds of of that operation. Additionally, we wanna look at, different methods with which that robot can can grab things and put them onto the pallet. So, I think we all understand that you can grab one tote and stack it onto a pallet. That's actually a pretty repeatable and pretty common process, but you can grab a row of totes and stack those onto a pallet. So in this case, based on the image I have on the screen here, we have, you know, one tote at a time on the bottom. And in the middle, so robot cells we have there, we have three totes at a time. So a row in this case would be three totes. It'd be a left row and a right row. And additionally, you can have a robotic arm move a layer of product at a time. This is a pretty popular methodology depending on the type of product you're moving. And and the reason this is popular is because the more things you pick up at a time, the faster that robot can hypothetically be. So let's say a cycle takes me thirty seconds, it might take thirty two seconds to move six of something where it takes me thirty seconds to move one of something. So, you know, a little bit of incremental increase in the, time to pick up that load, but a pretty substantial increase in the amount of totes per hour that a cell might be able to produce. So that's great. So we can increase the number of totes we move at a time. That can make our robot cells faster, and we could hypothetically have less of them. But at a certain point, you might need more robotic cells. And and why is that the case? So let's say that I have a robot cell that can put to one or two pallets at a time, or I have a robot sell, but I want to go real fast, so I need to give it rows of product or layers of product. But when I do that, I also imply that I have the same, destination totes all grouped together. So when I pick up a row, everything in that row needs to go to the same pallet. Some operations will have many, many destinations they need to sort to. So let's say I need to sort to forty destinations, but I really only have two or three or five robots to do that with. Well, and that might be absolutely the right amount of robots for the rate that you need. But, you know, three robots isn't gonna serve as forty different target pallets simultaneously. And, additionally, if you have random arrival of totes to a robot cell, you can't guarantee that they're going to arrive, in sets of three for one destination. So you have to do a little bit of magic before it gets to the robot cell. And here's where we get into robotic palatability systems. So we can do some stuff called buffering, and sequencing. So buffering is really just storing things in an interim kind of random access memory situation, until they're needed, or triggered by some other part of the process down the line. You're trying to soak up the slack for the variation between two different processes, and and buffers are a great way to do that. And sequencing, so this idea of sequencing is, changing the order of arrival of things. So maybe in in this image we have, on the left side, we have, totes arriving in one, two, three, four, five order. But on the right side of that robot, we want all ones to arrive first, and then all twos to arrive second, and then all three. So we're changing the sequence from, you know, a rainbow sequence to a very specific sequence that will allow that robot cell to do the most efficient work. Now there's a few ways to do this buffering and sequencing, at this kind of level. And and some, I think, we should talk about or or interesting to talk about it. There's conventional ASRS. So these are like mini loads, and we'll show what what one of those looks like in a moment. We have robotic ASRS systems. So in this particular image, we're showing some ACR bots and some racking systems. ACR is automatic case retrieval. So these robots will go in, grab cases, and move them around. You can achieve a similar effect with a buffer conveyor. So you can have a sufficient length of conveyor and a sorter, and you can do something pretty similar. Now if the tow count gets really high, that starts to become difficult and we'll kinda step through that as well. So this is an example of a, mini load buffer. Now in this example, what happens is, toast will come in kind of on the north side in a random order, and the crane will grab tote a tote or mini totes at a time and put them into its racking system. Those racking systems for many of those can be very tall. They can exhibit extremely high levels of product density. And oftentimes, you're using the store, products like case level operations where you need a lot of storage, but maybe you're not a hundred percent, sure that you need huge amounts of rate. Right? And you can double up aisles and increase rate, but you're gonna also have some more storage when you do that. There's a lot to consider here. But in this case, we're gonna feed this resources, buffering resource, and it's a mini load in this example, random totes. It's gonna put them away. And then while it's putting them away, it's gonna do an interleaved move. It's gonna take it's put away and combine it with a pick at the same time to save travel distance. So So it's gonna go in there and grab the toast that it's been requested for by the robot. So the robot's also saying, hey. Give me all the destination one toast. It's gonna put away some three fours and fives, and it's gonna go find several destination one totes and bring them down and put them on that south discharge line. And it's gonna go to the robot, and the robot's gonna build a nice little pallet, and we'll be happy. You can have multiple robots attached to that same discharge. So that's what the image on the left kind of displays. So here we have three robotic cells attached to the discharge of this mini load crane. Now we have a pretty great system. Right? We have three robots. They'll always get exactly the right sequence of, totes to build the pallets that they need to build. They could build any number of destinations provided they get the toast in the right sequence. And there's a ton of buffer. So they can work continuously ad infinitum really until that buffer's, exhausted. So you have some great machine utilization. You're also removing operation equation and and it really is the dream. Now one thing that would make this concept kinda challenging is that the crane is gonna be a bottleneck for this operation. So maybe you'd like to exploit if it was a faster system, let's say. Maybe you'd like to explore a type of buffering and sequencing solution that has some more higher rate characteristics than a single main load crane. But if it was really slow and you have lots and lots of destinations, this might be something, to consider. Now I'm gonna go back one slide, and we'll talk about this ACR style buffer systems. This is pretty interesting. So here, we can have a set of racking somehow in between the infeed conveyor and the robotic cell. Maybe we have a bypass in there. It's not really drawn because we're trying to keep it simple. But the robots will grab it, and the robots, you can have many, robots, and you can scale the throughput of that system by adding robots. Robots will grab it from the infeed line and stuff it in these little rack locations. And then when it's requested by any of the robots on the other end, they'll go grab grab the right totes for the robots request, and they'll take it to that cell and they'll feed it into the cell. So it does a really similar thing to the mini load example, but you can have a mini to mini relationship with the infeed line or lines, the robots that work the buffer so you can scale those up and down, and then the robots that are actually building the pallets. This is a could be a very high rate, system. Now maybe not the highest rate because you have to touch everything or you have to touch a lot of things twice since that is consideration for sure. But excellent flexibility, excellent scalability so you can start small and then add these chunks of rack or additional robots, over time incrementally. And it's definitely definitely something to consider. Now let's look at something that could be fast, but that can't necessarily handle tons and tons of variation. This is on the other end of the spectrum here. This is a conveyor system that will buffer, so it has a loop. It's like a recirculation loop. So things come in on the left side, and they they ride around this circle, until they're needed by the robot. So a conveyor system can buffer things to the extent that it is long enough to hold all the totes or the items that it's buffered. So this is kind of the catch with this type of system. But as soon as something is needed and it passes by a divert or passes by the robot so that it's needed for, it can go ahead and discharge, get to the robot cell, and you're you're good to go. Start building that pallet. This might be effective for systems that don't need huge amounts of buffer, but do need to move pretty fast. You're really only touching it in and then touching it again out when the robot picks it up versus the mini load example where you have to take it and put it into a rack and then take it out of the rack, bring it to a robot, etcetera, etcetera. So this this could be a a little bit better for those, but, the issue would be when you have large amounts of buffer. So if you wanna have lots and lots and lots of totes, in queue, this might not be a great solution. You might wanna look at a mini load or an ACR style buffer or something something a little bit more robust in its storage capacity. The solution, provided that we look at the data and all the math checks out could be one of the lower cost solutions of the set that I just talked about. Okay. So so now we talked about really simple palletizers. One or two pallets that's targeting. You don't have to have a big one arm. What if you wanted to target four or six or or more pallets at a time? So you could conceivably have less reliance on this buffering or or in feed system, but you're gonna have a more complicated robot cell at the same time. So here we're gonna increase the number of pal positions, and and we're not gonna have to buffer and sequence as much because we'll have more power positions, but as we'll talk, that's not always expressly true. The cells will be larger. The robot cells will be larger, and you'll have a greater degree of impact when you have to go into a cell and change out of pallets. We did talk about that a lot on the first one, but but when the robot finishes the pallet, that pallet has to be taken out, the full pallet taken out, and a new empty pallet put in so that robot can continue working. Now but before, when I did that, I kind of only impact one other pallet. In this case, I'm gonna impact the other five pallets in the six pallet system, that are operating. And if I have multiple robot cells inside of a single safety zone, I could potentially impact all of them. So it's it's something that you want. It's it's not always a bad thing, but it's definitely something to consider. And as the pallet count increases, and we've touched on this a little bit in the past, so too will the reach of the robot need to be longer. So now I don't have to just reach two positions. I have to reach four or six, and I'll have to go further and further, away from that infeed point with the, you know, the longest distance that I have to put tote. And this typically, it results in a heavier robot because that arms have to the arm has to be longer. Longer arms, more material, more materials, bigger motors, etcetera, etcetera. So it kind of results in these larger systems. And, kind of as in the previous methodology, we talked about inbound sequencing and trying to solve for those constraints. Here, we're gonna focus mostly on on the outbound sequence. So so what I mean in this case is, previously, we would use some kind of buffering and sequencing, feeding the robot cells to accommodate whatever changes in sort weight count, or or other issues we had. We would buffer and and sequence to the cells, to solve whatever the problem was, kind of a high level methodology note. In this example, what we're trying to do is have that buffer be more in the robot cells themselves. So we'll have more pallets available at any point in time. So I don't have to store up all my totes for pallet or for for destination number two. I can go ahead and send them to a robot cell that has destination number two available. So less buffering on the infeed side, but, more buffering within the cells themselves. So what what's that look like? Let's say that you have to have forty destinations, serviced by your robotic system, but you don't wanna have any pallet not represented at any time. So in a simple single system, we we would hold all the the totes for destinations that weren't being worked right now inside of a buffer. And here, we're just gonna send them to a robot cell that has that. That means that we have to have all of those pallet positions available all the time. So this is a fairly simple in feed system, but it's a little bit more complicated robotic cells. Right? If we have six of them, they're bigger robots, and and each of them has to be have sufficient rate to accommodate whatever load is generated by having whatever mix of six destinations assigned to it. So it becomes a pretty big system or it can become a pretty big system. It can take up a lot more space. It can it can be a lot higher cost. So here, in order to have forty destinations, I have to have one, two, three, four, five, six bots available to me. Whereas in the simpler system, I really only needed three simpler bots to handle forty or any number of destinations. Now that there's a lot of a lot of trade offs you make when you do that, but this is this is one thing to definitely consider. The other note on on something like this, and and we'll get into it a little bit later, is the high potential to overbuy capacity. So when you are designing a system, you wanna look at how many things do I have to sort to. So how many destinations? And let's say it's forty. And then how fast do they need to build totes to those forty destinations? Well, if we say it's two thousand an hour and we we build a system with six cells and it happens to be three thousand an an hour capacity, then I bought a thousand totes an hour capacity of robotics too much. Right? Alternatively, if I if I need some lower number or some different mix, the odds of me having a a misalignment between how fast I need all the robot cells to go in aggregate and the number of destinations I need to service with all those robots. It's it's hard for me to make those two things line up every time in an arrangement like this. So we talked about this kind of briefly, but, depending on how the pallet exchanging works, you may want to build in a bunch of extra space for machine access. So here in this image, we show a couple different methods. So we have a robotic forklift coming in there. We have a person with a pallet jack going into a robot cell, and we have a an AMR, a fork AMR, so four to four style machine, going into a robot cell and changing out pallets. So it's definitely safety is a big consideration with robotics. Ideally, robotics are left in their kind of own environment unless they're a cobot type of thing. But in in this case, you'd have to go through a safety barrier. If that robot cell is still working when you pass that safety barrier, then it would have to probably operate at a reduced rate so that everything else in that cell is gonna suffer. Sometimes it'll actually stop the robot cell completely while doing a change out, and that's gonna result in some lower robotic utilization than maybe planned for initially. And then while a robot cell is slowed or down for changing pallets, you're gonna have to buffer any demand that comes to it at the same time. So those buffer lanes that are feeding the robot cells, depending on how you change out pallets, could end up being pretty long. So changing out pallet positions can be kind of painful, but are there some ways to change out pallet positions that can reduce the burden on that system? And I think that there are. So some things that we can explore if if that is the question are, pallet conveyor arrangements. So on the very top image, we have a pallet conveyor loop. So pallets ride around this conveyor, and the cases or the totes come to the robot on its the same, infeed line we've been talking about the whole time. And, you know, we'll stop the pallet in front of the robot for the appropriate, destination. It'll grab the tote, stack it on there. Alternatively, we have, the lower left image. We have pallet spur. So we'll give the robot three pallets to work on in this example, and it'll build pallets and maybe we'll only sequence those three to it. And when one of them is done, we're gonna automatically take it away and then automatically drive a new pallet into that position with the conveyor. So this is it's just like, you know, I gotta say just like it's it's like case conveyor, but very, very big. So pallet conveyor. It's, pretty tried and true technology, and it works. It's not terribly fast. So the robot now won't necessarily suffer the same safety ramifications as it would if a a human entered the cell, but you do need to account for time to change over these pallets because it wouldn't be instantaneous. And then in the lower right, what we have is a pallet ASRS crane or a stacker crane, they're often called. And in here, we're just gonna have a robot work in tandem with an ASRS. So, ASRs will drop the pallet in its pick or put position. A robot will fill those up. And then when one of those is complete, the ASRS crane will grab that pallet, put it away in a racking system or put it on a discharge conveyor, whatever the disposition needs to be, and then put a new empty pallet in that position. So, this is a pretty interesting, arrangement, and I've seen this a few times in my life, and it can be pretty quick. You get an excellent buffer kind of in a natively out of the system. So, you know, maybe going to the next downstream process or an outbound area. It can be pretty powerful. It's not necessarily very expensive. You have to build an entire crane. These are are fairly large projects, so definitely something to consider. When looking specifically at the pallet conveyor, delivery and takeaway, you you do probably need to be a little bit more creative on this. So the change over time, as we've mentioned before, can be long. The loops, if you build a loop, you need to be pretty careful about what pallets are sequencing in that loop or what's working in that loop. This video like a lots and lots of sort ways, this could be a great solution. However, you wanna be cognizant of the rate impact of constantly cycling through these. There's also potentially applications where the pallets go so slow that you can place toes on them in motion. It's not something super great to plan on unless it's like a very specific application. And then kind of additionally, we wanna talk about integrating a desired number of robots to a single pallet conveyor system. Right? So we again, we don't wanna overbuy capacity. So so can I put multiple robots on one set of? Absolutely. Yeah. You totally can. And so we'll talk about what that looks like with another example in a moment here. Yeah. And kind of one additional note on these systems is because it's already on a pallet conveyor, you can attach this the the discharge of this process, the finished tote, to all kinds of other systems pretty easily. So any ancillary process, like, maybe you're gonna put a lid or some some slip sheets, or you're gonna run it through a wrapper or a strapper, or you wanna, weigh and dim the pallet after it's done or or send it to some other system. Like, it's already on the pallet conveyor, so it'd be a pretty easy presentation to do. Now here's an example of of having that some kind of technology to buffer your pallets, so your destination kinda target pallets, but then ganging up multiple robotic cells on that same system. So I don't have three ASRS, cranes for three robotic cells. I have one crane that can service many robotic cells. So now I'm just sorting or sequencing the right totes to the right robotic cell based on the pallets that have in front of them or or present. And and this can be these can be dynamic destinations too. So I can work a couple of fast movers and I can rotate my slow move around. As long as the system sorting the totes into the robotic cells and the ASR system are communicating and orchestrated appropriately, it should be should be a okay. So in this case, we're able to buy the right number of robots because we have this ASR as a lot of this buffer. So we don't have to over buy in this case. I think this example is probably one of my favorites of the collection. This is technology that's fairly new, but it's also pretty popular in the market right now. So we have these floor or shelf based AMRs. So these little bots that go underneath something, you can grab a pallet and take it away. We can use this AMR system, the shelf level AMR system to come in and manage which pallets are available at a robotic cell at any point in time. So that's it. I in this image, I have a robotic cell that has seven put two positions on it. It's a pretty big cell. But I can swap out those pallets such that I can effectively service ten or forty or a hundred depending on the rate that I need to achieve for that robot cell. So, you can multiply the number of sort ways that are available to to be serviced by a single cell by having an efficient system to manage what is in those put positions. And in this case, we've done this with pallet conveyor in previous example or, the pallet crane and ASRS. In this case, we're approaching with another robotic solution. So it's a robotic AMRs. Pretty clever, pretty low infrastructure cost, and, probably a probably a pretty excellent solution to explore if you have a high pallet count and kind of that middling, throughput account. And similar to the pallet, conveyor discussion, because I already have a pallet being taken away from the system on piece of technology, I could probably go ahead and tie that technology to some ancillary process. You can run an AMR into a strapper or wrapper pretty easily. And that's it's kind of a great additional benefit of a solution like this. Now let's talk about something that's a little bit more exotic. So this is, a rendering of something called a gantry. So this gantry is this big long rails, let's say, on the x axis, and it has these big gantries that ride across in the y axis. And inside of that gantry, there's a little, robotic head. So that'll move in the z axis. So each of these blue units can travel the length of the system and access any of these pallets. And in this way, they can cover hundreds of pallets per robot. So, the buffer in this case is just internal to the robot cell. So whereas a robot a six axis robotic arm that accesses six or eight pallets is on the larger end of a system like that. Here, it's it's completely reasonable to assume that these robots could could access four hundred fifty pallets conveyor brings pallets in and and it builds onto the pallet conveyor, takes the pallet full completed pallets out of system. That's definitely an option. You could also have a system like this that's serviced by the AMRs that we showed in the previous example. So an AMR will drive in and set the pallet in in somewhere in this service grid. And then, the gantry heads will go ahead and pick off of that or, you know, build to one of those positions. And then the AMRs will be triggered that, hey. Take this take this load away, so we'll go take it away and put a new one in that place. So definitely definitely a lot of options on how these work. Here's a here's another example. We have a field where it it expressly shows the AMRs servicing that Gaintree field. And and you can have a single head or single Gaintree bot riding on those rails. You can have multiple, heads. Although past two, it does get pretty, pretty wild. So typically, the most you see is two in a single system. And the field can be expanded or contracted based on the number of sort ways that you need inside that field. So you you the the I think in this example, we have something like fifty to cover forty destinations. So it's kind of overkill. But the last example was, I think, four hundred and fifty ish, kind of target locations, in that field. So pretty powerful stuff. And then similar to the other robotic systems, you can you can do single case, which is what we showed in the last images, or you could do rows of cases, or you could do layers of cases provided you feed and present those totes or cases in the appropriate manner. The more things you pick up at a time, the faster your totes per hour will be just like the other robotic solutions. So that does present an excellent option for these type of systems. Now you'll often see a system like this used in a layer pick environment just because it's it's a fairly high capital kinda entry cost and you wanna make sure you get lots and lots of rate out of it all the time. And so they're very often in layer pick operations, but not exclusively. And then there's kind of one other technology, and this is a much, more traditional, I guess, is the best way to say it, technology. This is a unit load former, that's what I call it, but it's a more of mechanical processes of conveyor and alignment. Maybe you have a a case flipper that comes and pushes the case into a certain orientation. And they'll kind of build each layer on a plant to to just use a word there. So on a surface, and that surface will then be retracted in some way and allow that layer to be set on top of the a pallet or the previous layer that was built. So here, you build a layer on the surface, and then you set it on the pallet. You build the next layer, and you set that on the pallet. You build the next layer, you set that on the pallet. These have the same kinda cool ability to connect to ancillary systems at all of the the pallet conveyor based systems and AMR systems to have. So you can run this right into a wrapper. Some of these systems can even wrap while they're building the pallet. Super easy to integrate into stuff, and you can build tote, or pallet de stackers into them so they can automatically put the next pallet in there. The one note is that you do need to be pretty locked down on how you sequence to these systems. Oftentimes, you'll see them used in production facilities where they're building the same SKU and the same, layer pattern again and again and again, and they're pretty excellent for that. It doesn't mean you can't use them for other things, but you do need to kind of pay attention to your data and ensure that you're gonna have the right mix for this type of machine. Okay. So let's talk about some some weird behaviors we see. So I'm gonna use some costs. These are not validated numbers. These are just, some inputs to the model to kinda give us some tangible takeaways, to think about and to talk about. So what's some some things that, you know, as as as we put this together, we we sort of realized that we needed to kinda share a little bit more is when you change the characteristics of the system. So when you get different design requirements for a system, the technology you end up selecting is not gonna be the same. So in this case, we have on the left side a system that runs at four thousand total per hour, but only needs to service two destinations. And and what we see is that we probably tend towards the simpler one to one or two to one style robotic systems. I I don't need a lot of destinations, and so it doesn't doesn't make sense for me to have these big systems where I can serve lots, lots of destinations. I I just don't, you know, I I do need, a lot of robotic capacity. So in this case, I'm gonna buy the right number of I'm gonna buy the number of robots based on the number of totes per hour that I need to achieve. On the far right side, what we see is a lot of destinations. So a hundred destinations that we're gonna build, you know, with hundred target pallets we wanna build to at a time, but only a thousand totes per hour. So here, we we're not gonna go as fast. This is a quarter of the speed of the first example, but it's, you know, fifty times more locations. There's a lot of locations. And in these examples, systems that service lots and lots of destinations tend to shine. So gantries, in this case, like specifically the layered entry, and the road entry, they look excellent because I I'm gonna be able to hit all those destinations very easily. Right? And I don't have to make that machine that doesn't move super fast move super fast. Right? I don't need all that extra rate. I do need the destinations. And in the middle here, we have this two thousand test per hour and forty destinations we wanna target. So this is it's kind of an in between case, if you will. And what you see is the things that start making the most sense are these medium complexity robotic cells. So four and six cells, and we're gonna sort to them. We'll do a little bit of of work there, but they start to make the most sense. So depending on your design requirements, depending on the characteristics of the final operation that that you wanna have, the technology selection is gonna be different. And this is definitely something that, you know, we as integrators, try to take as much time as we can to understand everything about your operation and then really dial that in. One note is, we have this mismatch count here, and this is kind of a a little metric I made up, a bit of a rule of thumb, if you will. And and what I do is I compare how many robots, I would need for any of these systems to service the throughput and then how many robots I would need to get the right number of pallets to destinations, right, to target destinations and see what the mismatch is. So in this case with the two destinations, there's a very small mismatch between those two numbers. I only need, two destinations. Right? But I need a lot of robots to hit rate. So I'm only, like, three apart in absolute terms in these in these two. But here and here, right, I need a hundred destinations, but each robot only does one pallet. Well, I'm ninety eight off because I need to buy a hundred robots to hit a hundred destinations in this case. But I only probably need two robots to, accommodate the throughput that's required, this thousand tPA. So I'm ninety eight apart. I can you know, I have a mismatch of ninety eight robots between capacity throughput capacity and and target destinations. And and this this mismatch, if you kinda think about it that way, will tend tend to predict what type of solution is going to be the best. Whatever has the least mismatch is gonna end up yielding the lowest capital for that scenario oftentimes. Not a not a hundred percent true all the time, and there's definitely, a lot this model doesn't take into account, but it's definitely, kind of some food for thought here. It minimizes you overbuying and that overbuying problem leads to increased capital in many circumstances. Now if we do the same kind of analysis, but we kinda dial things in a little bit more and we auto change out palettes, then what happens is that this behavior changes pretty significantly. So let's say I'm using an AMR based solution to automatically swap out palettes themselves. What happens is I end up making every robot cell similar to a gantry that it can service as many destinations as it needs to service. Right? This is it's not universally true, but it it starts to look like these one to one or one to two robot cells can service, forty destinations or two destinations. So what happens is when I auto change out pallets, I end up correcting for most of these mismatch issues. And then I can use, you know, a more complicated exchanging system, but, it's much simpler and much less expensive robotic system, to to achieve that rate. So now I'm I'm getting the right amount of robots for the rate, and I'm I'm solving for the destination count, with my exchanging systems, and or my my buffering and sequencing systems if those are of sufficient size and scale. And and here's, like, kind of a scenario to really try to hammer the point home, but we have this two thousand and forty scenario in both cases on the left side. This is the traditional, like, pallets have to be swapped out the hard way, and the gantry system looks pretty great, in a in a layer, regime if I can achieve layers. But, alternatively, like, in normal, you know, operation with singles of rows, the one to six robot looks pretty good. And you'll see robotic cells like this out in the wild, just because of this behavior. But if I auto exchange, then suddenly, what looks great in a lot of cases is these simple, really fast robotic palletizing cells. Now you'll notice that as the number of destination increases for robotic cell, so too do we believe that the rate of you know, the throughput would decline, and that's sort of what we model in here. That's kind of a note, and that's that's why this sort of happens. One big note and kinda getting close to the end here, so thanks for sticking with me. We wanna when we're designing these systems, we wanna look at your design requirements. And so that means, we ask the data what the right targets are for us to design this. As we discussed earlier in the presentation, depending on your goals, how many how how much throughput do you want? How many destinations do you want? How much weight does everything weigh? Like, these impact the technology selection. It's not just different, like, positioning. It's we we might select completely different technologies, depending on different design requirements. And there's, you know, there's evidence by the the amount of equipment and different arrangements that are available in the market today. But let's focus in on one case here. So two thousand hours, forty destinations. If I did it with a simple kind of pallet building robot cells and then some inbound buffering, the simplest type of minimum buffering, then I'd want probably three robots, and I'd want a buffering loop here that would sequence into these robotic cells to be pretty robust in terms of its ability to store things. So I probably need to store, like, nine hundred and twenty totes or something in in this kind of arrangement. So that's not impossible, but it is a bit of a make the conveyor that it's not really shown here. And and I'm gonna be running these three robots pretty hard. So if I actually go in here and exchange a pallet, it is gonna potentially cause me some heartache. So I can accommodate for that that pallet change out by making these double pallet cells, which probably don't lose any rate, or I can just add a couple of robots and make my whole system more robust anyway. So couple couple pieces to think about there. If I were to solve this two thousand an hour and forty totes with a kind of a medium complexity robotic cell, these these six to ones, for example. It might look something like this. This order is really just, saying, hey. This tow goes to cell one. This tow goes to cell five. This tow goes to cell six. That's really all it's doing, and it holds some additional inventory for when these guys are doing changeovers. But that that's about it. But you're gonna have to have a safety fence. It's gonna be a pretty big system because you have to have so many robots, and each robot cell has so many pal positions. They they they begin to become they begin to become pretty large systems with a lot of capital and a lot of footprint. And then if we look at, you know, this kind of gantry options with streaming all of your totes to a large multiple robot or multiple pallet, system, so like a gantry. This, in certain circumstances, is pretty great. Now I will say that, in the the forty destination and two thousand an hour range, what what tends to happen, or what we've seen here is that I can service all the destinations in one field. That's fine. But I'd probably need, in order to make the rate work for these systems, I need to have two of these gantry heads, and they would have to work at a layer level in order to have the right amount of throughput. So this is gonna imply that I have enough buffering feeding these things to hold, all of the totes that I would need to make, to provide single layers to the gantries, at a time. I probably have to have something like two hundred, totes in waiting, so I can I can build up enough to make a layer to give the gantries pure layers? Now I will say that when this is full, AMR is gonna take it away, and so it's it's also very highly automated. This is, you know, minimal interaction, but definitely definitely an interesting solution. The lower the throughput, the more attractive these get. And here's, you know, an option with the AMR swapping of the kind of normal robotic cells. So, you know, totes are gonna flow in based on the general destination. So I I show seven pallets per cell, but you can probably have totes flow in pretty regularly for twelve or fifteen, you know, destinations and just let the robot manage, the pallet positions itself. So we'd have software that would coordinate all of this stuff or orchestrate if you will. What we often see, as well is that pallets are kind of normally distributed. So you'll have some fast destinations and some slow destinations. And it's probably the case that the fast destinations would always be present at the robot cells, and the slower destinations would be the ones you swap out. Maybe that tote in the line for a little bit longer while the robot brings over a slow destination and swaps out an old slow destination. So that would be kind of what that arrangement would look and feel like. And here are the minimal downtime because we're interacting with all this power exchange in a robotic fashion. So it'd be minimal downtime, and there'd be minimal buffering, and sequencing on the infeed, and you'd be able to automatically connect it to downstream automation. So pretty pretty excellent outcome, really. Thanks for learning about pallet building arrangements with me today. I know this is kind of an interesting topic and some parts of it could be complex. Sometimes we go through it pretty fast, kind of the nature of the beast. If you wanna learn more about it, check out High-tech Intralogistics, we'll probably have some links to our website and some additional content, wherever you found this video.