Community projects updates

There are a few important updates to community projects that we would like to share.

First of all I would like to thank Lars Lorentz Ludvigsen for his great efforts to keep the local training toolkit up to date. Throughout the year he has been the lead developer and maintainer of the suite of projects, originally created by Chris Rhodes and then built on by Alex Schultz. I also must highlight contributions from Richard Fan and Matt Camp who have also created their flavours of a local training suite that helped us get through the training.

Starting from yesterday we have merged Lars’ repositories into community ones to mark a start to unification of community efforts.

Project formerly known as “deepracer-for-dummies” has been renamed to “deepracer-for-cloud” and is now the central repository for configuration of local or cloud based training solutions. You will find the project on GitHub. Lars has prepared a great guide to setting the project up on GitHub Pages. Note that you can run the training locally, in AWS (multiple regions, including Spot Instances), GCP or Azure.

Project previously known as “deepracer” has been renamed to “deepracer-core” to clearly mark that it is a repository supporting numerous builds and configurations for running the training. You will find it on GitHub.

There has been a small but important update coming to deepracer-utils. Version 0.10 brings unofficial support for DeepRacer to boto3 and AWS CLI. I would like to thank Don Barber for sharing this solution in the AWS DeepRacer Workshop repository log analysis tool. To install the support you need to install deepracer-utils version 0.10 or higher from pypi (using pip or conda) and then run:

python -m deepracer install-cli

This installs a service definition file in .aws/models in your home folder which means you can then use it for both AWS CLI and boto3. To check that it works run

aws deepracer help

And you should see help for deepracer together with the list of commands. The documentation is rather limited but we will be gradually expanding it as we learn ourselves what great things we can do with this. For a taster I recommend you try running this to see your racing alias (my output supplied):

$ aws deepracer get-alias
{
"Alias": "Breadcentric"
}

Thank you for all the community members involved in developing, reviewing and testing the community projects.