From Pod Robots to Item-Level Automation

The important point is that Amazon’s core warehouse robot system was built to move shelving, not to handle each item as a software controlled object. That matters because pod based systems save picker walking, but they still center the workflow on big mobile racks, fixed pick stations, and human grab steps. Newer architectures are increasingly built around finer grained movement, faster slot access, and more automation at the item level, which can lower labor per order and shrink the amount of building needed.

• Kiva’s original breakthrough was goods to person picking. Amazon bought Kiva in 2012, then scaled drive units that bring whole pods of inventory to workers. Amazon still describes this pod flow in its fulfillment centers today, even as it adds newer robots around it.
  1 aboutamazon 2 aboutamazon 3 aboutamazon
• The tradeoff is flexibility versus granularity. Locus took the same anti walking idea, but with robots that move through normal aisles and meet workers at shelves, avoiding Kiva’s heavier warehouse retrofit. Cytronic is pushing one step further, toward automated storage grids and robotic item picking inside a tighter, more closed loop system.
  4 sacra 5 sacra
• That helps explain the real competitive split in warehouse automation. Symbotic sells massive custom buildouts for operators like Walmart. Locus rents robots as an operating expense. Cytronic is trying to sell fulfillment as a service, where the merchant buys cheaper order handling, not a robotics project.
  4 sacra 5 sacra

The direction of travel is from moving shelves, to moving carts, to moving individual items with fewer human touches. Incumbents with pod based fleets can layer on new robots, and Amazon is clearly doing that, but the biggest cost gains are likely to come from systems designed from the start around item level automation and narrower, highly repeatable ecommerce workflows.