公开(公告)号：US10761520B1

公开(公告)日：2020-09-01

申请号：US15950624

申请日：2018-04-11

Applicant: Amazon Technologies, Inc.

Inventor： Stephen Boone ,  Joshua Davis ,  Matthew Faso ,  Cameron Fulford ,  Deepanker Kocheta

IPC: G05B19/418 ,  B25J9/16 ,  B25J9/00 ,  B25J13/00

Abstract: Embodiments herein describe forming clusters of network connected orchestration components (referred to herein as “orchestrators”) and distributing the management of a plurality of work cells among the orchestrators. That is, each cluster can include a plurality of work cell orchestration nodes which are the compute resources used to host an orchestrator for managing the work cells. Each cluster can be assigned to manage a particular type or version of a work cell. Because managing a work cell may use only a small fraction of the compute resources of the orchestration nodes, each orchestration node can manage multiple work cells. The embodiments herein describe distributing the work cells amongst the orchestration nodes using a work cell table which permits the orchestration nodes to assert ownership over new work cells and enable automated failover in case one of the orchestration nodes fails.

公开(公告)号：US10816964B1

公开(公告)日：2020-10-27

申请号：US15950460

申请日：2018-04-11

Applicant: Amazon Technologies, Inc.

Inventor： Stephen Boone ,  Joshua Davis ,  David Levy

IPC: G05B23/02 ,  G05B19/41 ,  G05B19/418

Abstract: Embodiments herein describe a fault tolerant network connected orchestrator which can handle network outages or hardware resets in a work cell. In one embodiment, the orchestrator determines the next task to assign to the work cell depending on whether the previous task was successfully completed. However, a network outage or a hardware failure may prevent the orchestrator from receiving the results of the previous action from the work cell. In one embodiment, the orchestrator recovers from a communication error by requesting the current state of sensors. Using this information, the orchestrator can deduce or determine the current state of the work cell and determine the next task for the work cell. In this manner, the orchestrator is fault tolerant such that it can recover from communication errors.