AI Driving Olympics
The AI-DO serves to benchmark the state of the art of artificial intelligence in autonomous driving by providing standardized simulation and hardware environments for tasks related to multi-sensory perception and embodied AI.
The Duckietown Foundation is proud to host these semiannual competitions at prestigious machine learning and robotics conferences around the globe.
For an overview of the current version of the AI-DO, including the various Leagues and their associated tasks, please check out https://driving-olympics.ai/.
Information about the Urban Driving League can be found here!
Urban Driving League
Urban driving is a cornerstone of autonomous mobility and navigating through these dense environments is challenging.
This league comprises embodied task. The environment is the scaled-down urban environment defined by the Duckietown appearance specifications.
Challenges for this league have included: Lane Following (LF), Lane Following with other Vehicles and obstacles (LFV), Lane Following with Intersections (LFI) and Lane Following with Vehicles and Intersections (LFVI). These challenges continue to be refined through successive AI-DOs.
After a competitors submission to a challenge is evaluated in simulation, the code is tested on physical robots in a Duckietown Autolab (robotarium).
Check out this video that describes what the Urban Driving League of AI-DO is all about!
(all times 9am EDT)
Thurs. Oct 31
- Description of the challenges
- Minimal agent template
Fri. Nov 1
- Classic template (ROS)
- Duckietown baseline
Mon. Nov 4
- Tensorflow template
- IL logs
- IL simulator
Tues. Nov. 5
- Pytorch template
Wed. Nov. 6
- Training in the cloud
Thurs. Nov 14
- Running on the Duckiebot
The first edition of the AI-DO took place in December 2018, at Neural Information Processing Systems (NeurIPS), the premiere machine learning conference, in Montréal. The was the first competition with real robots to take place at NeurIPS. AI-DO 1 only had the “Urban” league, and only the “Lane Following” challenge. There were over 1600 submissions from 58 unique participants.
The second edition of AI-DO took place at the 2019 International Conference on Robotics and Automation (ICRA), with finals held in Montréal, Canada, in May 2019. AI-DO 2 comprised again only the urban league, but additional challenges were added “Lane following with other vehicles” (LFV) and “Lane following with other vehicles and intersections” (LFVI). The number of submissions to AI-DO 2 was similar to the number for AI-DO 1.
The third edition was held at NeurIPS 2019 with finals held in Vancouver,Canada. In AI-DO 3 we introduced the advanced perception and racing leagues. In total we had over 2000 submissions across all of the leagues.
The fourth edition was scheduled for ICRA 2020 in Paris but has unfortunately been cancelled as a result of the COVID-19 outbreak. More details to follow about the time and location for AI-DO4!
Quotes from the experts
“The AI Driving Olympics is a great way to push the limits of deep learning on physically embodied systems.”
— Yoshua Bengio, University of Montreal
“The AI Driving Olympics offer a glimpse of the challenges of creating self-driving cars. A playful but rigorous competition on a smaller and safer testbed is the best way to develop the creativity needed to make progress in this field.”
— Emilio Frazzoli, ETH Zurich / nuTonomy
“It is great to see Duckietown host the AI Driving Olympics at ICRA and NIPS. What a fun way to demonstrate the real challenges in building and deploying self driving cars!”
— John Leonard, MIT
“Starting from MIT CSAIL, Duckietown has grown into a global initiative that is inspiring students around the world to learn about self-driving cars, as well as the science and engineering of autonomy.”
— Daniela Rus, MIT