In the past warehouse operations have typically hinged on two roles, pickers and packers. Pickers would travel throughout the warehouse, selecting goods that had been ordered by a customer. Those goods would be given to a packer, who would then prepare the order for shipping. These two components would work together, with the picker insuring the packer had a steady stream of the right goods to be packed for shipment. Kiva Systems, based in Massachusetts, has developed a robotic-based system that entirely transfers the role of the pickers to robots leaving the packers as the sole human element.
The Kiva system is simultaneously incredibly simple and dauntingly complex. Hundreds of orange robots the size of suitcases are built to transport carefully designed modular shelving units. These shelving units are organized into groups 2 wide and 5 deep, creating what is effectively a street system of "blocks" with open "streets" running in between them. The robots can drive under a shelving unit, lift it up over them (so the footprint remains the same), then carry the shelving unit over to where the packers are packing. A number of algorithms coordinate the movement of robots so there are no crashes, and designates what goods are on what shelving units so the robots can bring the right shelving units to the packers. At that point the packer just grabs what they need off the shelf right in front of them, after which the robot brings the shelving unit back to the "human exclusion zone" where are the shelves are organized.
The robots cover an average of 12 miles a day and only need to spend 5 minutes out of every hour charging. Amazon was so impressed by this technology and the possible impact it could have that they bought out Kiva Systems in March 2012 for over 700 million dollars. Since then most of Kiva's work has been exclusively with Amazon, as they work to refine their systems in a proprietary manner.
While Amazon has found a use for it in their fulfillment networks, the question is what is the next step for this sort of technology? While this system may work in a massive warehouse with hundreds of thousands of SKUs, does it scale down to smaller operation? A break-even point definitely exists where it becomes less costly to use employee pickers than robotic pickers, but the exact location of that point is difficult to locate. Lacking sufficient background or data, I imagine its on the larger end of the spectrum, as the fixed up-front costs of retooling an entire warehouse and purchasing the robot units needed is definitely costly and not to be taken lightly.
Another question to ask is if the picker-half of the picker-packer duo can be automated, is there a way to move in a direction where the entire warehouse is automated? In some industries this is easier to do, especially when the product is uniform. My father used work at a plant that made bricks, and while as a company they were technological dinosaurs, he took me to look at their process and it was easy to see how much of the work in their warehousing operations could be executed by robots since the dimensions of the objects being shipped were so similar. For Amazon, where they need to pack everything from jackets to computers to chairs, it may simply be too difficult to develop robotic packers that can handle all those different sorts of objects. One solution may be to develop specialized robots for different types of goods (one robot only packs clothes, another only packs small electronics, etc.) but programming them to handle every single possible SKU is likely beyond the realm of feasibility. With specialized robots it also would be difficult to adapt to changing orders, as bottlenecks may form (new fashion line comes out so everyone orders clothes, but one clothing packing robot would be quickly overwhelmed) but in this case human packers who can adapt are able to handle every type of good that comes their way.
While the future of Kiva's robotic picking fleet remains to be seen, it will certainly be interesting. A huge number of companies stand to possibly benefit by implementing the same or similar technology. My question to you is what direction should Amazon take with Kiva technology? Should they spin it out and make it available to the market as a business solution (similar to how their cloud services evolved?) or should they keep it proprietary to maximize their business advantage? Should they look to explore if humans in the packing process can be replaced like they in the picking process, or is that a futile effort and a waste of money?
Or should they look to form the worlds first robotic ballet company? Who knows what direction Amazon will choose to go with Kiva.
Additional links, sources, and resources: