Welcome to the DeepRacer Wiki

This wiki is designed to consolidate general information about the AWS DeepRacer into a cohesive wiki.

What is the AWS DeepRacer?

According the AWS website:

AWS DeepRacer is the fastest way to get rolling with machine learning, literally. Get hands-on with a fully autonomous 1/18th scale race car driven by reinforcement learning, 3D racing simulator, and global racing league.

Essentially, the DeepRacer is a platform to learn and refine Artificial Intelligence (AI) and Machine Learning (ML) techniques on an autonomous car. Unlike other solutions, such as DonkeyCars, DeepRacers can be fully trained and used in a virtual simulation. You don't even have to buy an actual hardware car. This lowers the bar of entry exponetially, and truly opens up DeepRacer to anyone.

Objective of this Wiki

• Capture the collective knowledge of the DeepRacer Slack Community
• Share that knowledge with others, to further everyone's join understanding & learning
• Grow the DeepRacer community and increase awareness
• Make information easy to access, to enable self sufficient resolution of common problems

Contributing

This wiki is maintained by the DeepRacing Slack Community and ARCC. This wiki was built using Docsify.

Install npm and use npm i -g docsify to install the docsify cli. Then run docsify serve ./ inside the repository to serve the webpage locally

To write new articles, simply create a new Markdown file in the root directory.

For example, the following page structure:

|- README.md
|- outline.md
|- getting-started
   |- README.md
   |- account.md

Would correlate with the these matching routes:

README.md                     => https://domain.com
outline.md                    => https://domain.com/outline
getting-started/README.md     => https://domain.com/getting-started
getting-started/account.md    => https://domain.com/getting-started/account

Headline Catagories

Headline categories:

• Getting Started
  • What is DR?
  • Why DR?
  • Navigating the knowledge base
  • ML overview
  • Reinforcement Learning
• The Community
  • Background / history / purpose
  • How to join
  • Who's who
  • Rules
  • Chat archive
• Training Technique
  • Overview of contributing aspects
  • Training process
  • Official Cloud vs Local Cloud vs Local Local
  • Reward functions
  • Action spaces
  • Hyperparameters
  • Execution (duration, tracks etc)
  • Notebook analysis
• Cloud training
  • Overview
  • Cloud training process
• Local Training
  • Overview
  • Concept & principles
  • The constituent parts
  • Getting going - the repos
  • Hardware requirements & compatibility
  • Local training process
  • Local training in the cloud!!
  • Problem solving
• Racing
  • Types of race
  • Virtual League Tips
    • Race track lessons
  • Physical races Tips
• Analysis
  • S3 File manifest
• Machine Learning Theory
• Meet-ups
  • Overview
  • List of locations
  • Upcoming events
  • How to run a Meetup
• Library
  • Cross links to useful resources elsewhere
• Getting help
  • FAQ
  • Ask the community on Slack
  • Live 'ask an expert' sessions
  • Archived 'ask an expert' sessions