Inspired by the success of AutoStore, a number of entrepreneurs and established vendors have developed their own variations of cube storage systems. In another article, I compared AutoStore to its competitor Jungheinrich PowerCube. At this point, AutoStore probably doesn’t have any serious competition in the cube storage space. While the AutoStore system isn’t “perfect” (with perfection depending on expectations), it’s certainly good enough for many purposes and applications. Perhaps, though, some of the features and enhancements of the competing systems can make a real difference. So, let’s take a look at it. In this article, I introduce and describe various cube storage systems and compare them to AutoStore as the benchmark for automated cube storage.
The Benchmark: AutoStore
AutoStore is both the name of the company and a material handling system developed on the rainy west coast of Norway in the 1990s. It was developed with the goal of maximizing storage density in an automated system. The system is based on a cube design where items are stored in individual bins within a grid structure. The bins are then accessed by a retrieval robot that moves horizontally on the grid and pulls up bins vertically within the grid as needed. The bins are then delivered to goods-to-person (GtP) stations for picking, of which there are several variants, all located adjacent to the storage cube.
AutoStore hasn’t only managed to very successfully market a system that many, including this author, initially thought made little sense (multi-deep storage for GtP!), but also has an excellent business model: sales and project implementation are handled by independent partners, of which there are now more than 20, essentially competing with the same system for the same projects – while AutoStore can sit back and count money. This is a brilliant move that no one else in the warehouse automation industry has pulled off.
It took a few years for AutoStore to gain a foothold in the market, but now business is booming, everyone knows (and often envies) them, they are profitable… Life is good.
Like any system, AutoStore has some limitations. I distinguish between limitations by choice and real pain points. Limitations by choice are, for example, the maximum net load of 30 kg per bin (more than sufficient for most companies) and the maximum overall system height of 6.300 mm. The pain points are inherent to the system design, and three pain points stand out:
- high sensitivity to floor evenness (because the bins stand on the floor),
- potentially tricky for fire safety permission (because stacked bins are hard to sprinkle effectively and authorities have been wary of cube storage since some Ocado warehouses have gone up in flames), and
- high sensitivity of retrieval performance to the ABC distribution of orderlines.
Let’s find out to what extent competing systems address these limitations. I will emphasize the systemic advantages the competing systems have over AutoStore, but will not address all the potential disadvantages (whether systemic or otherwise) in detail. Please keep this bias in mind while reading. The comparison table at the end of the article should provide more clarity.
Attabotics is a company based in Calgary, Canada. They call their cube storage system, including all components, “the Studio” – the storage cube itself is called “the Gallery”. At least that’s how I understand it. Apparently, the company has a penchant for flowery language. Another example? “Our insight-driven approach improves the flow of goods through rich data intelligence and AI/ML technologies.” – no offense, but of course that’s total bullshit.
Unlike AutoStore, but similar to Jungheinrich and Intellistore, Attabotics does not store bins directly on the floor, but keeps them in a rack, making the system less sensitive to uneven floors.
The system’s unique feature is that the robots can enter the cube through special chimneys (for lack of a better term). This is appealing because it allows direct access to all storage bins at all times – no digging required. If you think about it, this means that the system has more similarity with a tightly packed single-level shuttle system with double-deep storage and roaming functionality – just rolled over – than with AutoStore. But since the bots have direct access to four storage bins on each vertical level, it is actually better than double-deep storage. Effectively, it has a better storage density than double-deep shuttle systems while doing single-deep storage and retrieval. Not bad! But then again, storage density is lower than at AutoStore.
Since robots move both vertically and horizontally, the technical complexity of the machines should not be underestimated. Technical complexity is your enemy when it comes to production costs, reliability and maintenance, but I do not see any problems here that cannot be solved.
As for the system control, it seems to me that due to single-deep storage and full roaming functionality of all bots, storage and retrieval should be fairly straightforward, even when dealing with larger orders and sequencing requirements, such as is the case with online grocery. Also, because of the single-deep storage, I do not think the system should require any lead time before the retrieval and picking process can be started, making it more suitable for instant delivery (if that’s still relevant; the Q-Commerce folks are gradually going out of business). In contrast, AutoStore recommends allowing 30 minutes of lead time before picking so that the sequence of actions the bots perform can be determined and some bins can be dug out.
Attabotics has some sites in operation, but it is difficult to get an overview since they currently don’t feature references on their website. Their system at Nordstrom seems to have (had?) some problems. But hey, I have never seen a novel system without teething trouble, so wouldn’t judge them too harshly because of that.
Attabotics advantages over AutoStore:
- Much lower dependence on floor slab quality
- Much lower dependence on ABC distribution of orderlines
- Much lower process lead time
The Blue Robot Company is a small and relatively young (founded 2018) company from the south-east of Norway. The company is not widely known yet, and they don’t seem to have installed too many systems either (as suggested by the publicly available financial data), which doesn’t mean anything as they are just starting. It takes time to develop and market a system successfully, and everybody who dares to take on such an adventure deserves respect and support. There is one sales partner, Inge S. Årstad, a supplier of warehouse and workshop equipment from the Stavanger region at the west coast of Norway. They announced the installation of the system at a nearby customer, Norganic based in Klepp Stasjon, so we can conclude there is at least one customer. Blue Robot is the only company mentioned in this article that I never talked to, so I have to rely on what is shown in the videos and on the pictures.
As in AutoStore, bins are accessed from the top. What makes their system special is that they use gantry robots, not bots, for bin storage and retrieval.
What I see in the pictures on the vendor’s website are stacks of nine or ten bins. The required building height for small systems is 3.8 m.
The decision to use gantry robots is, of course, an important one. This is because it removes any possibility of adding additional storage and retrieval capacity to the system. You could add an entire system, but expanding dynamic capacity within the confines of the same cube is not possible. A rendered video shows two gantry robots with overlapping workspaces, but that’s probably all you can do in terms of dynamic capacity.
The public pictures suggest that bins could leave the systems, which is not the case for AutoStore. And AutoStore presumably has good reasons not to allow this: any damage to a bin can lead to an unbalanced stack.
The solution is based on Utz Eurotec bins. Three different heights are available: 220 mm, 280 mm, and 325 mm. The inner dimensions of the bin (length and width) are 566 mm x 366 mm, which I do not think will be a big problem for them (they are unlikely to store food retail cases in their bins which often have outer dimensions of 600 mm x 400 mm), but is certainly an unnecessary restriction. Most MHE companies these days use bins with inner dimensions slightly larger than 600 mm x 400 mm.
Unless I am missing something, there is no difference to AutoStore in terms of sensitivity to the floor slab or sensitivity to the ABC distribution of orderlines.
The website clearly states that the system is aimed at small and medium sized businesses. It could be argued that as AutoStore sells larger and larger systems, there may be a market opportunity with a cube storage system for small businesses – something even simpler than AutoStore. AutoStore has also recognised this and launched Pio.
Blue Robot advantages over AutoStore: none
Intellistore was founded in 2022 and is part of the Dutch OBS Group. Unlike Blue Robot and Template, they can therefore draw on the existing infrastructure and experience (and funding) of a more mature company.
The bins are stored in a rack (similar to PowerCube and Attabotics), which means they do not stand directly on the floor. This makes the system insensitive to uneven floors. It also enables Intellistore to do a magic trick:
Intellistore uses three different types of robots:
- A pusher robot moves under the storage bins and pushes up the stack from which a bin is needed.
- A buffer robot grabs the stack that was pushed up by the pusher robot and holds it
- The picker robot pulls the required bin out of the stack while the buffer robot holds it
This seems unnecessarily complicated at first glance, but at second glance it is easy to see that this logic eliminates one of AutoStore’s systemic pain points, namely its high sensitivity to the ABC distribution of orderlines. And although more different types of bots are needed, you probably need far fewer bots in total than in an AutoStore system to achieve the same dynamic performance. At least, this should be the case in principle since no digging is required (makes sense, doesn’t it?).
Indeed, this seems to be the most intelligent evolution of the basic AutoStore cube storage concept that I have come across. You can have your cake and eat it, too.
As for fire protection, there is a difference between AutoStore and Intellistore. Because the pusher bots move underneath the grid, a potential source of fire is underneath the bins, making it difficult to detect and extinguish. To extinguish the fire, injecting CO2 is Intellistore’s default response at this point. In my estimation, authority approval to use sprinklers instead is unlikely.
Three bin heights are available: 160 mm, 320 mm and 480 mm. With the lowest bins, stacks of 40 bins can be formed; with 320 mm high bins, the stack height can be 20. I have not witnessed the picking process live, but I would imagine that bins with a height of 480 mm are an ergonomic challenge for workers and should therefore be avoided.
Intellistore advantages over AutoStore:
- Much lower dependence on floor slab quality
- Much lower dependence on ABC distribution of orderlines
- Lower process lead time
I discussed PowerCube and compared it with AutoStore in great detail in another article, so I’ll be brief in this one: PowerCube can build higher systems and lift heavier bins. Technically, it looks like an upside-down AutoStore: the bots move below the storage grid. The fact that the bins are held by the grid makes the system much less sensitive to the quality of the floor. The bots can retrieve two bins at a time, but only one of them can be delivered to the GtP station in one cycle. The upside-down character of PowerCube allows the system to have a non-flat top: The height of the bin stack can be adjusted to the shape of the ceiling.
PowerCube advantages over AutoStore:
- Much lower dependence on floor slab quality
- Higher net load of bins
- Higher stacks of bins
- Potentially better utilization of building height below uneven roof
Template Systems is a start-up from Berlin. They are one of the youngest of the companies mentioned in this article. Conceptually, the system is an upside-down AutoStore, so it looks similar to PowerCube. The bots move underneath the stacks of bins. Unlike PowerCube, bots only store or retrieve one bin at a time, and they are much lighter. Like PowerCube, the Template system stores bins in the grid rather than on the floor, eliminating the high sensitivity to floor unevenness.
Similar to Blue Robot Company, Template targets a market segment below typical AutoStore installations; businesses with less available storage space, even lower ceiling heights, lower budgets… As mentioned earlier, AutoStore is trying to bridge the gap between manual operations and automation with their own Pio system.
At the moment, the company is still in the pilot phase and so it might be too early to include Template’s system in the full comparison as several parameters might change. I do hope they will gain a foothold in the market and we will hear more from them soon.
I have already published a review of Volume Dive, so I will be brief in this article.
Volume is a startup based in Dresden, Germany.
Like AutoStore, Dive’s bin handling robots (called “Snapper”) access storage bins from above. The key difference is that Dive has more than one level of robots in a cube. You can think of it as several relatively shallow (three or five bins deep) AutoStore systems stacked on top of each other multiple times.
In this way, Dive addresses several system limitations of AutoStore:
- Dive is very insensitive to the shape of the ABC distribution
- It is very insensitive to orders with many orderlines
- It requires less lead time for orders than AutoStore does
- The bins are held by the rack, i.e. the system is less dependent on the quality of the floor plates
Another point worth noting is that Volume Dive can work with “any” storage bin that meets certain requirements, meaning it does not require specific storage bins provided by Volume. The bins must have a footprint of either 600 mm x 400 mm or 300 mm x 400 mm, but the height is more flexible. It would have been better to make the bins slightly larger, as most of the other companies mentioned in this article do, but that’s something they can fix in a later iteration of the product development.
The downside to having multiple layers of robots is the splitting up of capacity: capacity planning (both in engineering the system and in its operation) will be more tricky. This situation is similar to other AS/RS, e.g., miniloads and shuttle systems (it is much worse for shuttle systems).
Conceptually, Volume does a lot right with Dive. I’m sure some details will change after the first projects, but the ideas are promising and I’m curious to see how the first projects will develop.
General Notes: Fire Safety and Maintenance
With cube storage systems, fire safety is a tricky issue. As you might imagine, plastic bins can burn (maybe there will be some with flame retardants someday?), and high-density plastic bin storage can burn like hell. UK online supermarket company Ocado, which has its own variant of a cube storage system, has had two major incidents in Andover (2019) and Erith (2021). In the Ocado system, as in the Autostore system, bots move across on top of the grid. Unless a battery ignites in a storage container, the bots are the most likely cause of fires. It’s one thing if the source of the fire is on top of the material. The heat will activate the sprinkler system and it will do its job. It is quite another thing if the source of the fire is under the storage cube. In that case, the fire, and therefore the heat source, is too far away for the sprinkler system to activate – until the fire gets really big and becomes a real problem. And then the sprinkler system does not work all that fantastically anyway, because the surface sprinkler may not reach the bottom of the cube. So bots that are below the storage cube make (a) fire detection and (b) sprinklering more difficult, if not somewhat useless.
An appropriate response to this is the use of Oxyreduct systems, and this is probably why Jungheinrich chose to work with the Wagner Group from the beginning to offer Oxyreduct as the preferred method of fire protection for their PowerCube system. In Oxyreduct systems, the oxygen content of the air is reduced to 13,5% – a level at which most materials cannot burn. So even if you were to bring a fire into the system, the plastic containers and their contents would not catch fire. I saw an impressive video of a fire test of two cube storage systems, one with surface sprinklers and one in a reduced oxygen atmosphere, and the difference could not be greater.
As for maintenance, for obvious reasons it’s a little harder to get to the bots under the cube than it is to get to the bots on top of the grid, which is presumably less convenient for maintenance personnel. The response I received from the companies with bots under the cube was that the bots drive out into the maintenance positions from underneath the cube. And if the bots, for whatever reason, die below the grid? Then someone drives in with a little cart. “It can be done”. I think it will be easier to find someone in your warehouse to pull an AutoStore bot out from somewhere on the grid to clear the way for the other bots to continue working, than to pull a bot out from the rather narrow space between the floor and the storage cube. This could be a factor affecting system downtime.
In this article I have tried to give an overview of the systems currently available on the market for automated cube storage. Autostore is far ahead of all competitors in terms of technological (and organizational) maturity, financial stability, number of references, sales process and maintenance network. Bots and software have improved over time and are considered reliable. The system availability is indeed remarkable; every AutoStore customer I have ever spoken with reported that the system virtually “never” goes down, which in addition to the technical concept makes system performance quite predictable, as long as order patterns remain stable. Yes – as long as the order patterns remain stable, but that’s the key, and that’s one of the weaknesses of the AutoStore system. Because if customers suddenly have a hankering for items that are normally slow in demand, but for whatever reason are now high in demand, the system falls on its face. This sensitivity to ABC distribution of orderlines is inherent in the system, and there’s not much AutoStore can do about it unless they change the principal setup. Which is exactly what some of their younger competitors are doing. As the article shows, some of the innovations are actually quite attractive. Attabotic and Intellistore address the problem of sensitivity to ABC distribution in different ways, but equally effectively, with Intellistore able to maintain the compactness of cube storage, while Attabotics gives up some of the compactness for the chimney solution.
Unrestricted access to all storage locations, without the need to dig, not only solves the sensitivity to ABC distribution of orderlines, but also the lead time required for order picking. This means that Attabotics and Intellistore seem to be better suited for express orders (for same-day or instant delivery). However, if you ask me as a logistics engineer, express ordering is overrated outside the world of time-critical spare parts supply, and it seems increasingly clear that most people are not willing to spend a lot of money on instant order fulfillment either.
Sensitivity to floor unevenness is another issue that bothers AutoStore customers, as it often results in expensive floor remediation in brownfield sites. Some late entrants solve this problem by storing bins in racks rather than directly on the floor. This, of course, makes the system more expensive, but the floor remediation also incurs costs that are just not as clearly visible in the integrators’ quotation documents. I wonder if AutoStore could not just build a low platform that mounts on the floor in the installation area, allowing them to use their normal concept without the need for racks while eliminating the effects of floor unevenness. After all, they do build the system on mezzanines (e.g. at TTI in Germany) several meters above the ground, so why not 20 cm high and eliminating the need for floor remediation?
I have not mentioned Ocado, partly because the system is pretty much the same conceptually as AutoStore, and partly because you do not just buy an Ocado system, you have to get married to their “end-to-end Ocado Smart Platform” which takes care of much more than just picking. Also, Ocado (currently) focuses exclusively on online grocery, which makes sense because they understand this business well.
To wrap it up: Some of the late entrants will be interesting to watch, and I would guess that some are here to stay. The warehouse automation market is big enough for many more companies to thrive, provided they offer something that actually delivers tangible benefits, i.e. they deliver systems that are either better or cheaper than existing solutions, or both. Which some of the companies featured in this article clearly do.
The following table sums up the comparison of some of the relevant features. Please note that + means “more” or “higher”, not “better”, just like – means “less”, not “worse”. I added green (for “better”) and red (for “worse”) to support interpretation.
After the publication of this article, other cube storage systems were officially revealed. Without going into details, I would like to list them briefly so that the reader can do his own research.
- Gridstore (https://www.gridstore.com/en/): Gridstore is a German company and, at the time of writing, is run by a former employee of SSI Schaefer. It looks like a cross between AutoStore and Intellistore with (optionally) two different types of vehicles running on the grid above the storage totes. “Ace” is the regular lifter robot and simply looks like an AutoStore robot. “Switch”, the second robot, can buffer up to three totes (but can only move two – for reasons I do not yet understand). The stated purpose of “Switch” is to assist “Ace” in freeing up totes that are stored in the lower levels of the cube. With other words: Bins can be dug out faster. All other things equal, this is an important benefit over AutoStore.
- Gebhardt Upstream (https://gebhardt-group.com/en/products/warehouse-technology-systems/warehouse-robotics/gebhardt-upstream.html): Gebhardt is an established supplier of warehouse automation technology based in Sinsheim, Germany. Of all the companies mentioned in this article, it is the most established and “oldest,” and it has an extensive portfolio of different solutions. At Logimat 2023, they introduced Upstream, a cross between a shuttle warehouse and a cube storage warehouse. It’s a single-level shuttle system with horizontal roaming (their Gebhardt OLS X Shuttle) with AutoStore-like lifting bots on the grid that replace the lifts that would normally be responsible for vertical tote transfer. Because the heart of the system is shuttle-based, all totes are stored in a rack and, unlike AutoStore, are not stacked on top of each other. The concept should eliminate much of the impact of the shape of the ABC distribution on system performance, which of course comes at the price of lower storage density. Since the lifting bots are responsible for delivering the storage totes to the pick stations, picking stations would benefit from an elevated position on a mezzanine floor to reduce cycle time (this is not shown in the promotional videos, but it would make sense).
- Instock (https://instock.com/en/): Instock is the youngest market participant. It is a Californian startup company, and accordingly things look a bit crazy. Just watch the promotional animation videos and you will know what I mean. The totes are stacked, but the stacks are low (the stacks shown in the video are only three totes high), but then the stacks of totes are placed on different storage levels in a rack. The retrieval robot is a small vehicle that can carry totes on its back (like a turtle AGV), but it can also drive up walls and drive bottom up under the ceiling of the different storage levels. When doing so, the vehicles transform into lifting robots that can place totes on stacks or lift totes from other vehicles. You might be interested in my conversation with Instock co-founder and CEO Yegor Anchyshkin.
- DenseStorage i-collector (https://www.densestorage.nl): DenseStorage is a Dutch company, and unlike the others mentioned in this article, they did not just launch the i-collector system recently – I just had not heard of them until very recently (thanks, Frederik!). There are several variations of the system, but essentially you can think of i-collector as a horizontal AutoStore-meets-carrussel-meets-sorter system. The multi-deep storage channels for totes (up to 24 totes deep) are approached from two sides by either vertical gantry robots or shuttles. If time permits, the totes can be stored in sequence, making it a neat solution for a shipping buffer because retrieval becomes really fast.
Many thanks to Marco Gebhardt, Yegor Anchyshkin and Manfred Welling for the information and the time they contributed to the addendum!
Did you know?
We also offer AutoStore consulting, planning, and analysis. Have a look at our services.
- 2023-02-23: Added section “General Notes: Fire Safety and Maintenance”. Many thanks to the gentleman who brought to my attention that these aspects should be mentioned.
- 2023-02-24: Minor revision
- 2023-05-11: Addendum added
- 2023-08-22: Minor additions
- 2023-09-29: Volume Dive added to review