A New Approach to Materials Handling in Warehouses

In a discussion in the TPS + 1 SENSEI group on LinkedIn, Casey Ng drew my attention to a materials handling approach from Kiva Systems, a company started up in Boston in 2003 by engineer/MBA Mick Mountz with funding from Bain Capital, that is now a wholly-owned subsidiary of Amazon. The following is a promotional video from Youtube:

The system shown in the video is clever, and can certainly be useful in fulfilling Amazon orders or in kit picking for assembly, but it is also obviously not a panacea. It only supports a single-level of racks, and boxes or bins that can be lifted by people. If you wanted to use the overhead space for storage, you might combine it with a classical automatic storage and retrieval system, which would move the portable racks to and from from upper levels for further handling on the ground by the Kiva pods.

Bringing materials to an operator at a fixed location rather than have the operator travel to do the picking is what is also attempted by carousels, but carousels require the operator to wait up to a half-turn for the right slot to be presented, and are limited in the number of items they can carry.

Cheesecake-factory-device-in-Kiva-warehouse with highlightAs shown in the video, while the concept is innovative in terms of storage and retrieval, it does not stretch hardware technology.  As we see on the video, we see the operator who loads boxes onto racks for putaway use devices that look like the ones used in restaurants like The Cheesecake Factory or the Fish Market to notify waiting customers that their tables are ready. The pods look like giant Roombas, but move in a more restricted manner. According to Mick Mountz, the pods just move around the grid of small squares marked on the floor with optical guidance and a simple form of “after-you” system to avoid collisions. On the video, auto-ID seems to be based on plain old barcodes. There is no mention of RFID or even QR-codes. The actual transfer of boxes is manual, with a form of pick-to-light guidance. While less visible, the software that coordinates all the moving parts is clearly at the core of this system.

I learned of Kiva’s existence this morning, and have no relationship with this company.


3 comments on “A New Approach to Materials Handling in Warehouses

  1. Comment on LinkedIn:

    I watched this and other Kiva videos on their process. In this one, they said employees could pick 2 or 3X orders per hour company. What I noticed was that the robots moved material to an associate who pulled the products and put them in different boxes BY order which were then transferred to the shipping area for final packing and shipping. That is fine if you have multiple line orders, but what if you have single line orders?

    If you have single line orders you can configure your order dispatch system to first segregate all single line orders and then aggregate the single line orders by SKU. The robot pulls the single SKU and delivers it directly to the packing and shipping station. In this case the packer pulls the product for each order by the ordered quantity. Since there is only a single SKU involved, the trip to the intermediate sort station is eliminated as is the intermediate handling.

    I have found very few companies that have ever done a frequency distribution of their orders by order size. I call this as great example of “Small Data” and simply looks at existing data by order size with counts, totals and cumulative counts by order size. In one company I worked with 19% of orders were for a single unit. Changing the picking to this process had huge productivity gains. In another, 1/3 of the assemble-to-order products were for a single line (could be multiple quantities), but still one line so no sorting/accumulation required. Sending directly to packing took multiple steps and one day out of the process.

Leave a Reply

Your email address will not be published. Required fields are marked *