You are currently viewing A Note on Goods-to-Robot Picking

A Note on Goods-to-Robot Picking


In the past couple of years, robotic piece picking solutions have experienced strong growth, driven by e-commerce operations and enabled by significant improvements in vision, algorithms, and gripping technology. Companies like RightHand Robotics and Kuka have enjoyed strong media coverage. Dematic and Righthand Robotics erected a first online grocery fulfillment center with fully automated pick stations for Drakes Supermarkets in Australia. TGW, Knapp and SSI Schäfer use all media outlets and tradeshows to showcase their robotic piece picking solutions. In every project I have had with eGrocery companies, at some point the conversation shifted to the question of maturity and business case of piece picking robots.

While good progress was made towards reliable and economically beneficial robotic piece picking, nobody asked: are we solving the right problem? And if we take a step back and look at it, we may indeed find that robotic piece picking in most instances is a great example of inside the box thinking.

Productivity as Driver of Automation

There are many reasons to (fully or partially) automate intralogistics processes. They include better ergonomics for warehouse staff, reduced risk of theft and breakage, and scarcity of labor in some parts of the world. In most cases, however, the dominant expectation towards warehouse automation is a good return on investment, and this translates into the need for massively increased productivity per warehouse worker. The biggest gains can be made by automating activities which have terrible productivity in manual operations. Examples are transportation of small loads (hence there are hundred thousands of kilometers of conveyors out there, and maybe soon equally many AMRs), storage of large amounts of goods (hence the existence of high bay warehouses and AS/RS) and lifting and stacking of boxes (hence the invention of palletizing robots and dolly stackers). Productivity considerations also underlie the conventional wisdom in warehouse automation that slow-movers be automated first: for fast-movers, manual picking productivity tends to be much less of a problem while it is often really low for slow-movers. Slow-movers thus tend to be the products that you store in Miniload or shuttle systems that feed goods-to-person stations.

Fixing the Non-broken Part

Talking about goods-to-person stations: Almost the entire logistics automation industry involved in the robotic piece picking discussion focuses on small articulated robots that would replace the human worker at the goods-to-person station, turning it into a goods-to-robot station. In a well-designed goods-to-person station, a human picker can easily sustain 400 to 500 picks per hour over a long period of time; peaking at up to 800 picks per hour is not uncommon. So the significant CAPEX spend on the goods-to-person solution enables pickers to achieve fantastic productivity (a word of caution: You should not make the mistake of omitting other process steps in your productivity calculation which are necessary to make the goods-to-person picking work, such as repacking into standardized plastic containers). And everybody seems to be frantically working on robots to replace these super-efficient human pickers, the most productive workers to be found in the entire warehouse. Also, the robots are slower (for now, and for some more years, but this will certainly change; don’t let robot providers fool you into thinking that a picking rate of 1.000 picks per hour can be sustained over any longer period of time), they are less reliable, so manual backup processes are necessary, they cannot normally cover the full SKU range, so complementary human picking is necessary, and they require maintenance. They don’t take coffee breaks, but they certainly do take maintenance and error reset breaks. This is not to say that goods-to-robot stations don’t ever make sense — they clearly can. But it takes systems of significant scale and the right circumstances in order for us to create a good business case by fixing the non-broken part, because we very much think inside the box.

Leaving the Box

Goods-to-person systems, and goods-to-robot stations alike, are normally powered by shuttle systems (and increasingly often by AutoStore type of hive-robot systems). As a rule of thumb, the cost of shuttle systems are approximately €1M per aisle. So we happily spend €10M – €40M – a common range for goods-to-person systems – and then we spend more time and effort trying to figure out how to replace the poor dude at the goods-to-person station whose total labor cost amounts to €40k p.a. Again, the business case can be positive (though this is certainly not self-evident). But how about trying to avoid building the expensive shuttle system in the first place?

Very often, there is no good alternative to a goods-to-person system with shuttles in: the building may already exist and its footprint may be rather limited, or the number of SKUs may be so high that only automated storage and retrieval makes sense. But often enough, this is not the case. Often enough, we are talking about less than 3.000 to 6.000 SKUs in total, all of which packed in boxes small enough to fit into conventional carton flow racks with tight separation of channels. Often enough, we want to increase productivity of hitherto purely manual operations in existing buildings. Rather than talking customers into shuttle systems and GtP, a well-designed zone picking system will do, and it is likely going to provide a better ROI than the good-to-person system.

Now – what about the robots? The articulated robots currently favored by much of the industry are not suitable for such systems. Their range is rather limited and they will face constraints picking from boxes in shelves; you would have to use rather large models with 2m reach and more, not comparable with the cute little robot arms you can find at goods-to-robot stations.

There is much to gain in such much less expensive (factor 10) systems, however, and it is time we are having a good conversation about common sense improvements rather than blindly repeating mantras about presumed benefits of goods-to-robot stations from corporate press releases.

Reach out to us if you are interested in discussing sensible applications of robotics in intralogistics projects, of if you are uncertain about recommendations received elsewhere.

Leave a Reply