EN
返回档案库

案例库 · 软件与 IT · 技术决策 · 2017-2022

这条还没译成中文,下面是英文原文。

Amazon's new picking algorithm cut robot travel 62% and saved $500M

Amazon redesigned which items robots fetch first; drive travel per pick fell 62% and the savings reached half a billion dollars.

Amazon

那一手

Amazon Robotics centres pair thousands of robot drives with human pickers: a drive fetches a storage pod and brings it to an associate, who picks the units an order needs. With millions of pods and orders, the picking algorithm decides which units to pick from which pods, and every poor choice adds drive travel.

Amazon redesigned that algorithm to minimise the distance drives travel per unit picked, then tested it in production. The new algorithm cut drive travel per pick by 62% with no negative operational impact, reduced the number of drives required by 31%, and was rolled out to every Amazon Robotics fulfilment centre.

The savings from fewer drives alone amounted to about half a billion dollars, and the approach also shrank the storage footprint by roughly 29% versus non-robotic centres. Amazon made AR fulfilment centres the standard for new buildings, betting the redesigned algorithm on top of the robots.

为什么管用

  • A pure software change deployed to every existing centre, no new hardware.
  • The metric was simple: drive distance per unit picked.
  • Fewer drives meant capital and energy savings, not just speed.
  • Robots and humans each did what they are best at.
值了多少Change the pick order, not the warehouse聪明

可以搬走什么

When the expensive resource is movement, the highest-leverage change can be the sequence of decisions, not the hardware. A smarter order of operations ships in software to every site at once.

后来呢

The algorithm became the standard for Amazon Robotics fulfilment centres worldwide; the case paper appeared in the INFORMS Journal on Applied Analytics in 2023 and was covered widely in the technology press.

资料来源

发现哪里写错了?告诉我们。

同一路聪明