Skip to content
An example of AWS Greengrass used in a miniature fulfillment center.
Branch: master
Clone or download
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
docs grammar tweaks Nov 16, 2017
groups bridge almost working Nov 17, 2017
models adding in base-to-femur servo mount spacer Jul 7, 2017
.gitignore
LICENSE Initial commit May 31, 2017
README.md Removed reference to manual CA python file. Dec 12, 2017
requirements.txt updated gg_group_setup version Nov 16, 2017

README.md

The Greengrass mini-fulfillment center example

At re:Invent 2016 AWS Greengrass was introduced to the world. This miniature fulfillment center demonstration was used to display what could be done with Greengrass when announced. It has since been upgraded to use the publicly available version of Greengrass.

A video was taken of the completed and operational Greengrass-based miniature fulfillment center, a.k.a. "dual robot arm demo" in action at re:Invent. Since re:Invent this demo and example has continued to evolve. It has been the seed for other demos and now supports Greengrass discovery capabilities.

The artifacts, example code, and instructions to construct this demo are available in this repository as an example. We look forward to seeing what you'll do with it.

Directory Structure and Noteworthy Files

aws-greengrass-mini-fulfillment/
  +- docs/
  |  +- BOM.md         -- the bill of materials necessary for construction
  |  +- CONSTR.md      -- instructions for physical construction (TODO)
  |  +- INSTALL.md     -- instructions for software provisioning and install
  |  +- OPERATE.md     -- instructions for operation of the demo
  |  +- TIDBITS.md     -- tiny pieces of setup information that might be helpful
  |  +- TROUBLE.md     -- troubleshooting of install and operation
  |
  +- groups/
  |  +- cert_setup.py  -- generates local certificates used by Greengrass groups 
  |  +- group_setup.py -- provisions and deploys example Greengrass groups 
  |  +- servo_setup.py -- prepares the servo communication code for example use  
  |  |
  |  +- arm/           -- the code and config necessary for an arm host
  |  |  +- ggd/        -- the Greengrass Devices that will be run on the Host
  |  |  |  +- servo/   -- the servo communications libraries 
  |  |  +- inv_arm/    -- the specific config and certs for the `inv_arm` group 
  |  |  +- sort_arm/   -- the specific config and certs for the `sort_arm` group
  |  | 
  |  +- lambda/        -- the Lambda functions used in the Greengrass groups
  |  |
  |  +- master/        -- the directory of the `master` Greengrass group 
  |     +- certs/      -- the group AWS IoT and local certs to be copied to Core
  |     +- ggd/        -- the Greengrass Devices that will be run on the Host
  |        +- certs/   -- the client certs used by GG devices
  |        +- flask/   -- where the web GG device gets files to serve
  |        +- servo/   -- the servo communications libraries 
  |
  +- models/           -- the models used to 3D print components of this demo
  +- LICENSE
  +- README.md -- this file
  +- requirements.txt

After following the installation instructions, each directory contains all the files necessary to support one of the three Demo groups' hosts. There are three Greengrass Groups named:

Group Host Name Description
inv_arm inv_arm-pi contains the Inventory Arm devices and the Inventory Arm Core
master master-pi contains both the Conveyor Belt devices and the Master Core
sort_arm sort_arm-pi contains the Sorting Arm devices and the Sort Arm Core

The demo behavior is broken into the concept of a series of simple control stages organized by Group. Those stages are:

Group Control Stages
Inventory Arm ... > Home > Finding > Pickup > Shipping > Home > ...
Master ..conveyor direction..
Sorting Arm ... > Home > Finding > Pickup > Sorting > Home > ...

As shown in the video above, even with these simple stages the interaction between the three Groups displays complex, cohesive, and coordinated behaviors.

This is possible because the whole system is modeled as a collection of micro-services. Those micro-services then leverage high-speed local messaging to enable communication within and between groups, as well as transactional state control using local device shadows.

The high-level, local architecture of the demo is: mini-fulfillment demo architecture

Master/Conveyor Host

There are six processes on this host:

  • the Master Greengrass Core
  • the heartbeat Greengrass Device (GGD) process
  • the bridge GGD process
  • the belt GGD process -- manages the conveyor belt servo
  • the button GGD process -- manages the button box's red, green, and white control buttons
  • the web GGD process

Within the Master Core, these are the Lambda functions:

  • MasterErrorDetector -- this function monitors local telemetry and detects any error states
  • MasterBrain -- this function monitors the fabric of telemetry coming from the bridges and the local GGDs. This function will decide what to do with respect to current telemetry and errors in the system

There is a web visualization on this host. If one browses to: <master_ip>:5000 a visualization that reads data from the Master Core can be seen.

Inventory Arm Host

There are three processes on this host:

  • the Inventory Arm Greengrass Core
  • the heartbeat Greengrass Device (GGD) process
  • the arm GGD process

Within the Inventory Arm Core, this is the Lambda function:

  • ArmErrorDetector -- this function monitors local telemetry and detects any error states

Sorting Arm Host

There are three processes on this host:

  • the Sorting Arm Greengrass Core
  • the heartbeat Greengrass Device (GGD) process
  • the arm GGD process

Within the Sort Arm Core, this is the Lambda function:

  • ArmErrorDetector -- this function monitors local telemetry and detects any error states

Noteworthy Files

There are some noteworthy files in the repository that centralize a fair amount of work around the instantiation of this demonstration.

  • groups/group_setup.py – performs all provisioning of each host's Greengrass Group. This file has commands that are useful when instantiating this demo and that demonstrate the use of the Greengrass REST APIs.
    • create <group_type> <config_file> – creates a Greengrass Group for a given group type and configuration file. Also stores all of the provisioning artifacts (i.e. GroupID, CoreDefinitionId, etc.) in the local configuration file.
    • deploy <config_file> – deploys a previously provisioned Greengrass Group
    • clean-all <config_file> – cleans up the entirety of a Greengrass Group and the locally stored provisioning artifacts.
  • groups/servo_setup.py – prepares the servo manufacturer's communication library for use on a Raspberry Pi with the Raspbian OS.

DIY

If you'd like to build a copy of this demo yourself you'll want to read both the installation instructions and the construction instructions. Please share any feedback you have or questions you run into here as Issues.

Special Thanks

Brett Francis and Todd Varland would like to thank the following. Without their help and existence we simply would not have made this demo.

The Francis Family, The Varland Family, gg-group-setup, ROBOTIS, Servode, Pretty Lights, Sigur Ros, Flume - Insanity, Amazon.com, HBO Silicon Valley, The Greengrass Developers

You can’t perform that action at this time.