Warehouse automation has been around for many decades. It is a complex and expansive industry that has, in recent years (specifically the last five to ten), been characterized by a surge of innovations \u2014 or rather, a lot of \u201cnew stuff\u201d, some of which is genuinely innovative, while other is less so. I thought it might be interesting to categorize this \u201cnew stuff\u201d into different buckets to better understand what\u2019s happening in the market.<\/p>\n\n\n\n
When I teach classes at the university<\/a>, there are a few lessons I consider \u201ceternal truths.\u201d Some are domain-specific, while others are more general, often originating from outside academia. One such eternal truth is that while tools change, principles do not<\/em>. This concept profoundly impacted me when I first encountered it, and I believe I first heard it from Tim Ferriss (\u201cThe 4-Hour Work Week\u201d).<\/p>\n\n\n\n The realization that tools often change is both a blessing and a curse. On the one hand, it allows you to focus your efforts on developing skills that will not become obsolete \u2014 the blessing being that these skills remain useful for many years, possibly even a lifetime. On the other hand, learning to use a tool you suspect may become irrelevant can create significant internal resistance. For example, I had little interest in learning to use Qlik Sense and Qlik View when the data analysis department at my former employer adopted them. With the limited time and attention I had during my long and busy workdays, I was concerned that my knowledge and skills would become useless the moment we changed the tool or I changed companies (a concern that proved justified). These days, I have limited interest in mastering Power BI (for instance). Instead, I stick with Python for data analysis because, as a general-purpose language, it will likely be around for decades. Even if it is eventually \u201creplaced\u201d by another programming language, my Python skills can easily be transferred to almost any other language in the \u201cC\u201d family. I started with the basics of C when I was 13 or 14 years old. While I no longer need to worry about pointers and buffer overflows, the general structure and principles I learned back then still apply today when I work with Python, some 25 years later.<\/p>\n\n\n\n The idea that principles don\u2019t change also led me on a journey to distill planning principles for warehouse automation projects, which has become a central tenet of my industry seminars as well as the courses I teach to logistics engineering students at THWS. I found that planning engineers in the warehouse automation industry often learn their craft in an unsystematic way. If you were lucky, you had a mentor who not only knew a lot but was also capable of communicating and sharing that knowledge \u2014 two qualities that seem to rarely coexist in engineers. Most engineers I\u2019ve encountered were simply thrown into the deep end, assigned projects, and left to figure out how to manage them as they went along. This was my experience as well; I had no concept of how to approach a new project (or any project, for that matter).<\/p>\n\n\n\n But back to the buckets: the notion that tools change while principles don’t<\/em> gives us two clear buckets. The first bucket is tools<\/strong>. When I say tools, I mean technical systems that operate within the confines of a defined process and according to a specific principle. Examples of tools in the context of warehouse automation include:<\/p>\n\n\n\n The second bucket is principles<\/strong>. As far as technology is concerned, principles in warehouse automation might include:<\/p>\n\n\n\n This bucket requires the most creative effort and out-of-the-box thinking, and it is the one that receives the least attention. Changing principles, or developing new ones, can be a significant engineering challenge, but it doesn\u2019t have to be. Take zone picking, for example. The mechanical engineering involved in a conveyor-based zone picking system is not rocket science, but it took some bright minds to develop the concept. The control system, too, is not very complicated (in most systems that have been built). The same applies to goods-to-person picking. But coming up with these concepts in the first place and executing them initially deserves considerable respect.<\/p>\n\n\n\n The third bucket is incremental improvements<\/strong>. These are the changes \u2014 presumably improvements \u2014 applied to existing tools or categories of tools. Examples include a Miniload crane that becomes slightly lighter, a new shuttle designed for higher racks, or an AGV battery that lasts longer.<\/p>\n\n\n\n Sometimes, what appears to be the development of a new tool is actually an incremental improvement. Take a new AutoStore look-alike that can move larger or heavier bins: yes, it is a new AS\/RS with a new name from (perhaps) a new company, but it is essentially an incremental improvement on the original cube storage concept.<\/p>\n\n\n\n Incremental improvements generally require the least amount of creative effort. All the variables are defined (velocity, versatility, payload, cost, etc.), and all you do is try to improve any number of them. It is simple but certainly not easy. Incremental improvements can present significant engineering challenges.<\/p>\n\n\n\n Moreover, incremental improvements are at the heart of continuous improvement (Kaizen), which is widely known but frequently poorly implemented. Many companies have some type of continuous improvement process (CIP), but many of these initiatives lead to little real improvement because big-picture types in management are too busy to address the (seemingly) minor problems that are often the focus of CIP efforts. Without exception, every CIP I\u2019ve seen in action has stopped short of any true root-cause analysis, making them superficial and limiting their potential impact.<\/p>\n\n\n\n One warehouse automation company I\u2019m familiar with had a tendency to launch a new product every couple of years to replace an existing product that had flaws making it difficult to sell. Each new product came with teething troubles that led to project delays, significant IT implementation effort, marketing and training effort, and new flaws. Instead of abandoning existing products and replacing them with new ones, the company should have directed serious effort at addressing the flaws, which would have cost a fraction of what new product development and all the subsequent costs entailed. A proper CIP in this case could have made a significant difference and likely altered the company\u2019s trajectory quite drastically.<\/p>\n\n\n\n Now that we\u2019ve defined the buckets, let\u2019s consider where most change occurs and what that implies.<\/p>\n\n\n\n Overwhelmingly, the changes we hear about in the industry \u2014 at LogiMAT, ProMAT, Modex, on LinkedIn, and in press releases \u2014 concern tools. Companies introduce new robots, new AS\/RS, new AGVs. These tools operate according to established principles and within familiar processes. AGVs move load carriers around. AS\/RS store and retrieve load carriers. They support different purposes and processes (an AS\/RS could be used for raw material storage, spare parts storage, or to support a goods-to-person process or serve as a shipping buffer), but all these processes are well-defined and well-known.<\/p>\n\n\n\n We don\u2019t hear much about incremental improvements in warehouse automation, partly because they aren\u2019t as attractive for marketing purposes. Or at least that\u2019s what companies seem to think, not only in the warehouse automation industry but across many sectors. Moreover, incremental improvements often don\u2019t matter as much to the broader audience.<\/p>\n\n\n\n Truth be told, though, the new tools we\u2019re presented with don\u2019t matter much either. Yet another shuttle system, another cube storage system, another AGV\u2026 they perform the same functions as all the other systems. Perhaps they do so slightly differently, but the tasks remain unchanged. So much so that, as I mentioned earlier, some of the newly developed tools could be considered merely incremental changes to existing tools. Even AutoStore, when it was founded in 1996, did the same thing as everyone else: storing and retrieving totes. They did it differently, but the system\u2019s purpose was still automated storage and retrieval for goods-to-person picking.<\/p>\n\n\n\n The bucket that rarely experiences innovation is principles. Very few companies address principles, and when they do, most observers fail to notice or appreciate it because they don\u2019t understand what they\u2019re seeing. So, let\u2019s discuss some examples where innovation in the principles bucket has actually taken place.<\/p>\n\n\n\n Let\u2019s start with robotic picking. For the most part, robotic picking exemplifies inside-the-box thinking. Almost literally. Most robotic picking falls into the category of goods-to-robot picking, meaning you install a goods-to-person system and swap the person with a robot.<\/p>\n\n\n\n I\u2019m not very impressed with that idea for several reasons, which I explain in detail in this video<\/a>. You combine all the disadvantages of robots with all the disadvantages of AS\/RS-based storage. There are applications where it makes a lot of sense, but these are actually quite niche. The small installed base of goods-to-robot systems speaks for itself. The total investment (including internal investments at warehouse automation companies and private equity investments in robotics companies) poured into this kind of solution to date exceeds the project volume for the next two decades; RightHand Robotics\u2019 series C funding alone is probably sufficient to cover the next few years.<\/p>\n\n\n\n Robotic picking can be used in smarter ways, however. Consider British online supermarket\/tech company Ocado, which places its robotic picking arms on top of its storage grids<\/a>. This approach (a) saves space for pick stations, (b) shortens the travel distance for hive robots that bring storage totes to picking robots, (c) eliminates the time-consuming vertical transport of storage totes at pick stations, and (d) facilitates the consolidation of manually picked volumes at pick stations. This is a significant improvement over bringing storage totes to pick stations.<\/p>\n\n\n\n Another example of a company that is demonstrating out-of-the-box thinking (again, almost literally), is Pickr.ai<\/a>. Pickr is a small Norwegian robotics company that has gone one step further and, in their concepts, eliminate the need for an AS\/RS altogether. What they do isn\u2019t rocket science, but they have realized that having to install a hugely expensive automated storage and retrieval system in order to be able to do cost-efficient robotic picking eats up most of the benefits you could generate.\u00a0 They do not<\/em> build mobile<\/em> picking robots which in my estimation wouldn’t make a lot of sense. Instead, their gantry robots operate within small-scale rack structures. They can be used in zone picking systems or in autonomous picking cells. And yes, this does limit the range of SKUs they can physically access whereas goods-to-robot type of systems and Ocado\u2019s evolution of it can have access to all SKUs in a system. So, it\u2019s not quite comparable, but it doesn\u2019t have to be, because the investment needed is such a small fraction of that needed for the other systems that the business case will be vastly superior. It truly is a disruptive innovation in Christensen\u2019s terms. \u00a0<\/p>\n\n\n\n Outside robotic picking, I\u2019d also like to mention another company that has chosen to operate outside established principles: BionicHIVE<\/a>. By now, almost everyone has heard of them, but if you haven\u2019t, I\u2019d like to direct you to my podcast with them<\/a> as well as my follow-up article<\/a>. BionicHIVE builds AGVs that can climb racks. Conceptually, their closest cousin is Exotec, with the crucial difference being that Exotec requires custom-built racks for their Skypods, while BionicHIVE only needs a cheap add-on to existing racks. It\u2019s unclear to me if the team at BionicHIVE has fully figured out precisely what they want to achieve with their technology; it certainly won\u2019t be goods-to-person picking. However, it is evident that they could explore many opportunities beyond the picking realm that everyone else is so focused on.<\/p>\n\n\n\n I\u2019m highlighting these companies because they operate very differently from almost any other company. Instead of developing new tools that fulfill the same tasks as dozens of other existing tools on the market, they have redefined their tasks. Pickr.ai is considerably undercutting the cost of existing robotic piece-picking solutions because they figured out that by focusing on a relevant subset of the market and a relevant subset of orders to be picked, they save on the cost of AS\/RS (which represents the major chunk of costs in a robotic piece-picking system). BionicHIVE has decided they want to provide automation support in brownfields without creating massive projects. Ocado has moved their picking robots onto the grid because they realized that the concept of having robots operate in stations on the floor is based on human constraints, which don\u2019t apply to robots. These three companies have little in common, except for the fact that all of them have broken out of the rigidity of existing concepts and are demonstrating out-of-the-box thinking at a high level.<\/p>\n\n\n\n I have previously stated that I believe warehouse automation has reached a technological plateau<\/a>. This year, I went to LogiMAT in Stuttgart, expecting to see little to nothing that was truly novel. One reason for this expectation is that most of the \u201cnew stuff\u201d (I hesitate to call it innovations, as most new products and systems don\u2019t qualify as true innovations) falls into the tools <\/em>or incremental change<\/em> buckets. As long as this remains the case, I think warehouse engineering as a discipline will not change significantly. That\u2019s not necessarily a bad thing, but it\u2019s important to understand because we are unlikely to see big leaps forward or the lights-out facilities that have been promised for more than 30 years. And: no, AI is not going to change that anytime soon.<\/p>\n\n\n\n Moreover, I believe that warehouse automation still needs a lot of incremental improvements. I find it hard to accept that so many automation projects still finish behind schedule due to software problems. I also find it challenging to accept that certain automated systems, such as single-level shuttles, still come with enormous commissioning efforts, and that AGV projects, even after more than 40 years, still frequently encounter issues. But we don\u2019t necessarily need more or novel tools, we just need to get our act together.<\/p>\n\n\n\n It is also evident that many warehouses suffer from rather basic problems that can be resolved without significant (or any) investments in automation. In every single warehouse audit, we identify simple measures that significantly improve operations, so much so that warehouse expansion or automation projects can be scaled down or stopped altogether. Therefore, there is ample potential to be tapped without relying on advanced technology. So, here we don\u2019t need new tools, either. Instead, we need good understanding of data and what decisions we can derive from it. We need warehouse managers who are actively trying to improve operations. We need communication between warehousing and purchasing, marketing, sales, and production. Perhaps we need more audits<\/a> (I think so<\/a>). \u00a0And perhaps, breaking out of conventional operating principles to bring down the cost of automation will make a big difference here, too, but that\u2019s certainly not the first step.<\/p>\n\n\n\n Meaningful progress and sustainable competitive advantage in warehouse engineering comes from rethinking principles, not just refining tools. Innovative companies are not just tweaking existing systems; they are challenging the foundations of how we approach warehouse engineering. By stepping outside the conventional boundaries, they remind us that the future of this industry won’t be shaped by improvements in speed, capacity, or efficiency, but by principles that allow us redefine concepts. But that\u2019s hard and takes lots of creative effort.<\/p>\n\n\n\n\n
\n
Where Change Happens<\/h2>\n\n\n\n
Innovation in Warehouse Engineering Principles: Examples<\/h2>\n\n\n\n
Why This Matters<\/h2>\n\n\n\n
Conclusion<\/h2>\n\n\n\n