We are excited to announce the The AI Driving Olympics (AI-DO), a new competition focused around AI for self-driving cars. The competition uses the Duckietown platform which makes heavy use of ROS.
The AI-DO competition focuses on the hard problems of embodied AI, such as deploying learning on complex systems, choosing the right modularity of learning processes, the challenges of learning in simulation while deploying in reality.
AI-DO 1 is held in conjunction with NIPS 2018. The competition will take place in the cloud and in robotariums in Oct 1 - Dec 1, with a live final event on Dec 7 at NIPS.
AI-DO 2 will take place at ICRA 2019, with a timeline to be announced.
The competition comprises 4 challenges of increasing complexity: 1) Road following on an empty road; 2) Road following with obstacles; 3) Point to point navigation in a city network with other vehicles; and 4) Fleet planning for a full autonomous mobility on demand system.
Competitors will send in their code to be evaluated first in simulation, and then in Duckietown robotariums around the world. Acquiring a Duckiebot is not required but is possible through our kickstarter (bitly.com/DuckietownKickstarter).
The AI Driving Olympics is presented by the Duckietown Foundation in collaboration with 6 academic institutions: ETH Zürich (Switzerland), Université de Montréal (Canada), Tsinghua University (China), National Chiao Tung University (Taiwan), Toyota Technological Institute at Chicago (USA), and Georgia Tech (USA).
Our two industrial sponsors and organizers are nuTonomy and Amazon Web Services, which, in addition to financial support, are providing state-of-the-art infrastructure for the project.
To read more about AI-DO, please see http://AI-DO.duckietown.org.