GridPick: a Puzzle-Based Order Picking System

My graduate student Onur Uludag and I are happy to announce GridPick, a new semi-automated order picking concept for piece- or carton-picking. GridPick is based on the puzzle-architecture of GridFlow, but instead of receiving cartons from one side and delivering to the other, it moves cartons to and away from the same side—the pick face.

GridFlow is a new approach to the design of “fast pick areas” in distribution centers. In a typical application pallets of a product (or stock keeping unit, “sku”) are kept in a reserve area. Because it is inefficient to pick orders from these pallet locations, a separate fast pick area is created, in which smaller quantities of popular skus are kept. The idea is to increase the “face density” of skus (measured in skus per square foot of pick space, for example), so that workers don’t have to travel as far between picks.

A typical flow rack system

To the right is a flow rack, in which each lane contains a single sku. Lanes are typically 4-5 cartons or totes deep. When the lane is almost empty, a worker replenishes the lane from reserve stock. Flow rack has a face density of one sku per lane.

GridPick increases face density beyond one sku per slot by presenting only needed skus to the pick face, then withdrawing them to make room for upcoming picks. While the worker walks along the pick face picking the current order, items in the next order are making their way to the face in preparation for his next order. Ideally, workers never have to wait for an order not already at the face.

Below is an illustration of how GridPick might compare with an equivalent flow rack. Both systems have 5 pallets each of 48 skus, for a total of 240 pallets. The nature of flow rack obliges a length of 48 slots, whereas this implementation of GridPick (other aspect ratios could be considered) has 29 slots. This is an excellent illustration of GridPick’s higher face density.

The plot below shows how average productivity improves for a worker in a GridPick system. The level of benefit is higher for smaller orders because the ratio of travel time to pick time is greater in this case; because GridPick significantly reduces travel time, it offers more benefit for smaller batches.

Productivity increase of GridPick over an equivalent flow rack, for different average order sizes.

We will be posting more information about GridPick in coming months. We welcome your feedback in the comments section!


  1. Looks a little like an application for a Hattland Autostore solution : will be interesting to see the dynamic pick face reaction time !!

    1. Yes, we have thought a bit about the similarities between GridFlow/GridPick and AutoStore. If you imagine the top of AutoStore with many, many robots (and new control algorithms, of course), then they are very close.

      We will be posting a short paper on GridPick soon. Thanks for your interest.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s