Narrow-Aisle Warehouse Positioning Solutions | Marvelmind
Warehouse Indoor Positioning: Industrial Use Case
Narrow-aisle warehouses present unique challenges for indoor positioning systems. When aisles are long and confined, standard 2D tracking degrades because beacons form poor angles for trilateration. This practical guide reveals two proven solutions: mounting sensors above shelves for reliable 2D tracking, or deploying multiple 1D tracking submaps along each aisle with intelligent handover zones. Learn when to use each approach and how to optimize beacon placement for maximum accuracy.
Transcript
Narrow-aisle warehouses present unique challenges for indoor positioning systems. When aisles are long and confined, standard 2D tracking degrades because beacons form poor angles for trilateration. This practical guide reveals two proven solutions: mounting sensors above shelves for reliable 2D tracking, or deploying multiple 1D tracking submaps along each aisle with intelligent handover zones. Learn when to use each approach and how to optimize beacon placement for maximum accuracy.
0:03 Let's discuss today tracking in some special cases, for example in narrow aisles. So this photo we received today from the customer, and the customer asks us how to track in such a pretty complex area. And yes, the answer is not simple at all, because what is the problem? The problem is that this is very narrow and this is very long. So it means that if you typically install one mobile beacon here and another there, the ratio of the width to the length will be too small. So it means that the width is too small as compared to the length. So effectively, if you combine it with the height, it makes the trilateration even more problematic. And as
1:03 a result, instead of 2D tracking you have 1D tracking. Yeah, you can try to do 2D. It will be kind of working, and you can see the video. So watch this video. It will be kind of working, but the closer you come to the line connecting the stationary beacons, the worse it will look. So you will still have very accurate X tracking. If X is this axis, and the more you come to this area, the worse the Y will be. Okay, we can try to jump to let's now discuss the opposite scenario. X is so this is the situation. You will see that the X coordinate will still have plus minus 2 cm, but this will be having I don't know plus minus 10 cm, and the closer you come to the line, it will be plus minus 15, plus minus 20 cm. At some point in time you cannot talk about 2D
2:05 tracking anymore. And that was pure 2D tracking. In this case, there will be also height difference which will introduce additionally narrowing of the angles, and the trilateration will be poor. Okay, but the situation is very typical. What to do in this case? Well, one of the solutions is that you just track above the shelves. So for example, you have a robot, and the robot is a scanning robot. The robot drives between the shelves. It scans the QR codes. It kind of estimates what products are there and populate some database or CRM or warehouse management system. Um, so what you do, you have a robot, and then you have a you install on the robot a small pole which will be above the shelves like
3:06 in this case. So you install the stationary beacons on the ceiling, on the walls. It basically doesn't matter because, for example, this doesn't look like a huge warehouse and is very densely packed. So it meant that your robot must be above the shelves. Not the robot itself, but at least the microphones. They have a very easy to use 1-meter cable. So it means that your robot can be 1 meter below the microphone. The microphone will be above the top shelf, and the microphone will see the stationary beacons which I installed there. So in this way, you avoid all kinds of issues. It's a basic 2D tracking, and you can
4:07 see their placement manual. Check the 2D tracking recommendations. So this is your case. So you install the stationary beacon there. But instead of a small robot, whatever robot you have, a robot with, for example, this is the height of your shelves. So your robot which is I don't know 1 meter tall or 1.5 meter tall robot, and then you additionally install the pole, and the pole is above the shelf. For example, this is the height of the shelf, so the pole is above the shelf, and you have 2D tracking in this area, easy. So you avoid, since you sometimes, or not sometimes but quite often, you need not only the location but direction. So you install you install two mobile beacons with the base starting from 25, 30 cm. If you have a larger one, like half a meter for example, you will have a very accurate not only location but also direction. So that's the very typical case. So
5:10 the first solution is and the easiest and recommended: if you can do that, track above the shelves, and there will be no problem at all. It will be very basic 2D tracking. If you cannot do that for whatever reason, for example, you track a person or track a tool in a person's hand, then use 1D tracking in aisles. Let's look at the map in more details. So this is the shelf, this is shelf, this is shelf, this is shelf. And this is aisle. Very narrow aisle. For example, I don't know half a meter or 1 meter aisle. Very narrow. And this is, for example, 10 meters long. So what you do, you install one stationary beacon here, one stationary one stationary one stationary, and each of the stationary beacons they cover this area. They cannot cover through shelves, through shelves, because there will be no line of sight, and line of sight is a must. So it means that you create
6:16 multiple small submaps with a single stationary beacon 04, beacon 05, beacon 06, etc. And each of these beacons serves its own service zone. So this is service zone, this is service zone, this is service zone, etc. These two stationary beacons are taken in 2D, and you can see they can cover a huge area. For example, 25 by 25 meters. So the maximum distance would be less than 30 meters. Recommended, some cases could be more, but we don't recommend overstretch the capabilities. But this area, this what we are discussing, very narrow multiple small submaps. Of course there will be handover zones, so it means that when you move, for example there is a door, you come here. You are in 2D tracking, 2D, 2D, then you go to their handover zone 05A. So this is a handover zone between submap 01 and the sub
7:18 map 04. Then you go, and then you move to this handover zone which is a handover zone between submap 02A and 04. And you are still in 2D here. And then you jump—hopefully not jump, but soft handover—to 02A service zone. Of course, even in 2D you need the height of the mobile beacon. So you move, move, and when you move to 1D, the 1D tracking assumes that the height remains the same. So it's a big assumption, not too terrible assumption, but still. For example, you have people of different heights, or people are moving arms with the tool up and down. So yes, you will have some inaccuracy when your height will differ from the height that you enter, but it will not be too much difference. So it means that you move, move, you move from 2D to 1D, then you come to this area, then you return, etc., and you switch between 1D tracking to 2D tracking. Um, so that
8:28 that would be the recommendation, because it helps you save on the number of beacons. You don't install two beacons instead of one in this area. And second, even if you install two beacons, you still effectively have 1D tracking in this area because the width or the length is very long. At the same time, the aisle is very, very narrow. So there's no point to install the stationary beacons, and you know you save 50% in this area. This area is typical 2D. This area is typical 2D. This area is typical 2D. You can do 3D as well. But if you have only 2D tracking, there's no need to create 3D submaps. Again, you need more beacons. The accuracy of 2D will be still higher than in 3D, and you don't benefit. Um, what to pay attention to? Well, with
9:30 this 1D, it's clear. So you just have multiple submaps for each of their aisles. But some areas, for example, will be tracked less perfectly than some other areas. For example, this is the problematic area, this is the problematic area, this is the problematic area, and this is the problematic area. And this, for example, is a problematic area. And this and this. Okay, why is this the problematic area? Well, for the same reason. Look, so this is the beacon and this is the beacon. If you're in this area, in order to track you effectively have zero or whatever two degrees angle. So your triangle is almost 1D. So it's not a triangle anymore. It's a line. This means that the tracking in this area will be poor. So um, what other problematic areas? Uh, so this area is problematic for the same reason as this one, because these beacons cannot
10:40 serve them well, because it's very, very narrow angle, and these beacons cannot serve well because it's very narrow angle to them. So yeah, there will be tracking, but you can expect good tracking quality or accuracy in this direction and poor in this direction. So um, it would be possible for some very, very complex systems to combine this information with this information, to take X from this which will be Y in this area, to take Y from this which would be X in this area. But it's overly complicated. Just remember that in this area the tracking may be not perfect at all. In this area the tracking will be not perfect because of ultra-wide base. You see, it's too close. So it means that you will once again have a very good tracking in this direction and very poor quality tracking in this direction. So effectively
11:46 it will be, for the same reason as here, you will have 1D tracking in this area. So it means that we do not recommend to come too close to this. Typically it's not an issue in practice because you cannot come within whatever 20 cm from the shelf. You always keep whatever half a meter, and the tracking is already reasonable. But it's always not about the matter of distance but about the ratio between this. It means that if you want to come too close, or let's say very close, then the distance must be reduced. Not because of maximum distance limitation, but because of this ratio, this distance versus this distance limitation. Um, another area of problematic areas, this one, for example. Like in this, if you have a distance from this beacon more than 30 meters, we do not recommend to have it more than 30 meters. There's no clear cut, and if you have whatever 32 meters
12:47 it's okay. But don't overstretch, because the larger the distance, the higher chances of noise, the higher chances of missing the tracking in this area. So just remember about this. As you can see, this picture or this map, this map, and this map, slightly different. So what's the difference? The difference is when you have not only very narrow but very long aisles as well. What does mean long? It means that the distances here already are more than 25 or 30 meters. So it means that one beacon cannot cover the whole length of the aisle. Sometimes the aisles could be I don't know 50 meters. So that may be not enough. So what you do? Well, the simplest solution: you install one beacon here, as always or as before, and another beacon here, maybe around 25 meters. And the length
13:52 of the submap would be 30 meters, with this area to be a handover zone between submap 02A and 10A. And this is the handover zone between them. So this beacon starts effectively serving from this area. Oh actually, I need to correct myself. So this beacon must be here. So it cannot see behind it. So the beacon must be here, like 04. So the black point must be around or underneath this. This means that it will start serving here and it will continue to this area. Um, so this is just a variant of the same. No fundamental difference. All other areas are the same. But simply, when you have a larger or longer aisles, you will need also more beacons within the aisle to cover it. What are these beacons? Uh, these beacons, these beacons
14:57 can be regular Super-Beacons or Industrial Super-Beacons. We typically recommend to install them on magnetic holders. So it means that you can point to the right direction. Okay, it's here easy. But for example here, the beacon must not point like this. It must point to the center. So it's better to use the magnetic holder. But then again, when it refers to the center, then you may create additional issues in this area. For example, when you point to the center, then this beacon or this area may be even less covered. So there is a trade-off, but it's not a subject for this discussion because we are discussing how to cover these particular cases with narrow aisles. Um, some other things to remember: potentially problematic. So line of sight, line of sight revolves or defines everything, and everything revolves around line of sight. The problem is that if you install beacon like this, for example here and here, these beacons
16:01 cannot penetrate through the shelves because of multiple obstacles on the one hand or on the other hand. In some cases, when the shelves are empty, it can penetrate, and thus interfere with the system in the neighboring aisles. So it means that it's important to choose the frequencies wisely. And how to choose the frequencies, there's an article about the submaps, how to build submaps, what submaps are, and read those articles. Okay, my internet connection stops working anyway. So you can find them. You can find them here: basics about submaps, how to build submaps. Check them. It's important. And um, if you have very many small aisles,
17:09 ultrasound frequencies. Because there are only eight ultrasound frequencies: 19, 22, 25, 28, 31, 34, and 45 kHz in our system. In large warehouses, that's not a problem because you have larger distances. You have larger submaps, and the next time when you need to reuse the frequency, it's far enough: 40 meters, 50, 100 meters. So that's not a problem. But when you have such a densely packed area, it is a problem. Because their next frequency may be just whatever 20 meters away, and with 20 meters it can propagate through the empty shelves, for example, and then interfere. So you may need to use time division multiple access. But time division multiple access between. So it means that you don't have the same submap using the same ultrasound frequency in
18:09 Inverse Architecture too close. But if you use time division multiple access, so this is in one time slot, another time slot. So basically they don't coexist in the same time. So they don't interfere with each other by definition. But time division multiple access brings another problem, which is lower update rate per mobile beacon and some other issues with handover. So if if possible to reuse without TDMA, use without TDMA. Um, so if you are using Non-Inverse Architecture, instead of using more stationary beacons, you can use horns. So you install stationary beacon and a horn. And in this case, it can be not up to 30 meters but up to 50 meters or even 100 meters in some cases, if it's, for example, a low noise area. If there is a high noise, then you need to reduce the area to keep the signal-to-noise ratio high enough. But remember that you can install a mobile stationary
19:14 beacon with the horn and increase it. It works in Non-Inverse Architecture because the horn can receive the stationary beacon. It can receive the ultrasound. But if you use Inverse Architecture or IA, then the horn is useless unless you're using so-called TX horn. We don't have such a product, but we experimented with them, and we can offer it for larger cases. So, so TX horn where there's a transducer and the horn, not the microphone and the horn. So, um, again a bit more about the terminology. So 01, 03, 04, etc., they are service zones. Between them, 05, 05B, and 06, they are handover zones. These are the mobile beacons N02, N03. We use the same terminology as in Dashboard, and
20:14 uh as I already mentioned, the handover zones. These are the shelves, and the submap is a set of beacons. So in this case, N02 and N03 is a set of two beacons creating a 2D submap with a service zone 01. This is another submap: N1 and 12. This is a submap, and N09 and 10 is another submap. So these are 2D submaps, but these are 1D submaps. So N07 creates a submap, 2D with the service zone 02. So this is a submap, and this is a handover zone between this submap on this service zone and this service zone. Service zone is a property of a submap, and submap is a set of beacons. So um
21:25 what else to mention? Now I already mentioned about this and about 1D, 1.5D, etc. So it may look funny, but in reality it's a very, very real thing. So in this case you have 1D tracking, but you also need to know which shelf are you looking at: this shelf or this shelf? So this is why we call it 1.5D. So it's 1D plus direction, but the direction is provided mostly by the mobile beacons. So install two mobile beacons on your robot, for example, or on your forklift. And oh, you can see it here in the placement manual, as we already saw. So install two mobile beacons, and then you know the location and the direction. So where your mobile beacon is facing: to the left shelf or to the right shelf, for example, like in this case. Um, for people
22:33 it's more difficult. So this is why there's no solution serving all. So this is why we always ask specific questions. And those questions can be found here. So what are you tracking? What people? And what do you want to achieve: productivity or safety? We need a floor plan for you, photos, videos, etc. So um, so this is why some solutions, like for forklifts, is easy. For people is problematic, more problematic let's say it's also doable but more complex. And for drones, typically you don't care about this. You care not to kill the drone by touching with the propellers and crashing the drone. Okay, there was some technical interruption. So with 5.2
23:41 uh drone tracking is still possible. 1.5 is the same along the X-axis. It's used ultrasound. 0.5 is um by the mobile beacons, so direction of the drone. And Z coordinate for the drone is possible from a barometer, for example. So you have a height from the barometer. 2.5D tracking from the forklift is also possible. But instead of a barometer, you have Z tracking against the stationary beacon on the forklift itself. So of course, against the walls, the stationary beacon on their forklift is not stationary, it's moving. But for their mobile beacon on the fork, it's stationary, so you get the coordinate from there.
24:45 So it's typical, and we use the same approach for cranes tracking, for example. So if you want to see how it works in practice and handover is between 1D tracking, then 3D tracking in this area, then moving back 1D tracking, and then there will be 2D vertical tracking on the stairs. So you can watch this video. It's nothing unusual, and it just helps you to optimally use the equipment. So with few beacons you can achieve the same. So um, about this special areas, we already discussed. Remember, so this is a primary topic of this discussion. But there are some other areas, like this one. Oh sorry, like these ones and like these ones, which are available or which are present in many other cases. So we are just describing them in more details here.
25:50 So remember and enjoy, and remember that since we build the solution from scratch, so you can nearly always find an optimal configuration. If you are not sure, check the placement manual. If you are even less sure, then contact us over email, and we are always happy to help. Thank you.
Video Contents
Key Takeaways
- Narrow aisles cause unfavorable geometry that degrades 2D tracking into 1D, with Y-axis errors reaching ±15-20cm near beacon connection lines
- The recommended solution: mount mobile beacons on poles 1+ meter above shelves to achieve proper 2D triangulation with ceiling-mounted stationary beacons
- Deploy multiple independent 1D tracking submaps along each aisle connected by 2D handover zones to reduce beacon costs by 50% while maintaining positional accuracy
- 1D tracking assumes constant height, acceptable for robots and forklifts but less accurate when personnel lift tools or reach upward
- Identify problematic zones where beacon angles are nearly collinear and expect tracking degradation; optimize beacon placement to maximize triangulation angles
- Use two mobile beacons separated 25-50cm to track both position and direction heading in narrow-aisle environments
Relevant For: Warehouse Managers & Factory Automation Teams
Warehouse managers and automation engineers deploying autonomous robots, scanning systems, and forklifts in densely-packed facilities with narrow aisles. This content solves the critical challenge of achieving accurate indoor positioning when standard 2D tracking fails due to unfavorable geometry—where aisle width-to-length ratios create poor trilateration angles.
FAQ
Industrial Indoor Positioning System
Narrow-aisle warehouse environments demand specialized indoor positioning strategies. The core problem: when aisles are extremely long relative to their width, the geometry between stationary beacons creates shallow triangulation angles, degrading from true 2D tracking into effectively 1D tracking. This results in excellent accuracy along the aisle length but severe Y-axis errors (±10-20cm) and problematic height calculations. Marvelmind's field-proven solutions address this through two approaches. First: mount mobile beacons or microphones on extended poles above shelves, creating proper 2D geometry with ceiling-mounted stationary beacons—the recommended method for robots scanning QR codes between shelves. Second: deploy multiple independent 1D tracking submaps, each covering a single aisle section with one or two beacons per zone, connected by handover regions where 2D tracking temporarily activates. This submap architecture saves 50% on beacon hardware while maintaining position accuracy by exploiting the physical constraint that most motion occurs along the aisle axis. The presentation details problematic zones where narrow angles cause tracking degradation and explains why mixing 1D and 2D tracking modes, combined with fixed height assumptions, provides practical accuracy for autonomous forklifts and mobile robots despite confined geometry.
Topics
Related Resources
📍 Need precise indoor positioning for your project?