Home
build details

# The AI Driving Olympics

Modified 4 days ago by Liam Paull

## Welcome to AI Driving Olympics!

Modified yesterday by Kirsten Bowser

Despite recent breakthroughs, the ability of deep learning and reinforcement learning to out perform traditional approaches to control physically embodied robotic agents remains largely unproven. To help bridge this gap, we created the “AI Driving Olympics” (AI-DO), a competition with the objective of evaluating the state of the art in machine learning and artificial intelligence for mobile robotics. Based on the simple and well specified autonomous driving and navigation environment called “Duckietown,” AI-DO includes a series of tasks of increasing complexity – from simple lane-following to fleet management. For each task, we provide tools for competitors to use in the form of simulators, logs, code templates, baseline implementations and low-cost access to robotic hardware. We evaluate submissions in simulation online, on standardized hardware environments, and finally at the competition event.

Participants will not need to be physically present—they will just need to send their source code packaged as a Docker image. There will be qualifying rounds in simulation, similar to the recent DARPA Robotics Challenge, and we will make available the use of “robotariums,” which are facilities that allow remote experimentation in a reproducible setting. New this edition is onsite testing on the competition grounds in the days leading up to the final.

• AIDO 1 is in conjunction with NeurIPS Dec. 2018.

• AIDO 2 is in conjunction with ICRA May 2019.

• AIDO 3 is in conjunction with NeurIPS Dec 2019.

## How to use this documentation

Modified yesterday by Kirsten Bowser

If you would like to compete in the AI-DO, you will probably want to do something like:

At this point you are all setup, can make a submission, and you should want to make your submission better. To do this the following tools might prove useful:

• The AIDO API so that your workflow is efficient using our tools.
• The reference algorithms where we have implemented some different approaches to solve the challenges.

## How to get help

Modified yesterday by Kirsten Bowser

If you are stuck try one of the following things:

• Look through the contents of this documentation using the links on the left. Note that the “Parts” have many “Chapters” that you can see when you click on the Part title,
• Look at the questions page on the website and see if someone has asked the question that you have (and if not feel free to ask it),
• If you are sure you actually found a bug, file a github issue in the appropriate repo.

## The challenges server

Modified 2 days ago by Liam Paull

See the leaderboards and many other things at the site https://challenges.duckietown.org/.

## How to cite

Modified 2 days ago by Liam Paull

If you use the AI-DO platform in your work and want to cite it please use:

@misc{zilly2019ai,
title={The AI Driving Olympics at NeurIPS 2018},
author={Julian Zilly and Jacopo Tani and Breandan Considine and Bhairav Mehta and Andrea F. Daniele and Manfred Diaz and Gianmarco Bernasconi and Claudio Ruch and Jan Hakenberg and Florian Golemo and A. Kirsten Bowser and Matthew R. Walter and Ruslan Hristov and Sunil Mallya and Emilio Frazzoli and Andrea Censi and Liam Paull},
year={2019},
}