Announcing: the AWS DeepRacer Community race data repository

Today, we are happy to announce the release of the AWS DeepRacer Community race data repository. Over the past few years, we have seen thousands of racers compete in the AWS DeepRacer League, and together they produced tons of leaderboard submissions. During the past weeks, we’ve worked on archiving and collecting the track and leaderboard data which is available via the AWS DeepRacer service. The archived data can be found on Github and is updated every hour.

Currently, two resources are available: track data and leaderboard data.

Track data

Po-Chun Super Speedway, this month’s PRO division track.

The track dataset contains track metadata such as name, description, track length, and track width as shown in the AWS DeepRacer Console. The track dataset is updated daily at approximately 1 PM UTC. The following data is available:

  • Track metadata for each track available in the AWS DeepRacer console.
  • Assets for each track so you can get a quick overview.
  • A mapping of NumPy files used for log-analysis 

Currently there are 27 tracks for which we have data available.

Leaderboard data

The leaderboard dataset contains leaderboard metadata and a historical overview of all submissions. For each leaderboard, information such as name, description, participants, and specific configurations for head-2-head and object avoidance is available. 

For all races starting in 2021, we provide hourly updates on leaderboard standings in easy to process CSV files to gain insight into the rankings of all league participants. For older races, we archived the final leaderboards.

The following data is available:

  • Leaderboard metadata for each leaderboard.
  • Assets for each leaderboard (if available).
  • For races before 2021: the final standings in a CSV file.
  • For races during this season: hourly snapshots of leaderboards as CSV files.

We have data for 62 leaderboards, with submissions for more than 14000 unique racer aliases for you to explore.

We’ll continue to maintain and improve the dataset over time. If you want to learn more about AWS DeepRacer, and the racers found in the dataset, please join us at the AWS DeepRacer Community Slack channel.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.