CAMBRIDGE, Mass. and ZURICH and MONTREAL (PRWEB) August 10, 2018
The Duckietown Foundation is excited to announce the inaugural The AI Driving Olympics (AI-DO), a new competition focused around AI for self-driving cars. The first edition of the AI Driving Olympics 2018 will take place December 7, 2018, at Neural Information Processing Systems (NIPS), the premiere machine learning conference, in Montréal. This is the first competition with real robots that will take place at a machine learning conference.
AI-DO aims to address some of the critical scientific and societal challenges related to AI and robotics. It directs academic research towards the hard problems of embodied AI, such as modularity of learning processes, and learning in simulation while deploying in reality. It also further promotes the democratization of AI and robotics research by using an inexpensive platform, and offering a common infrastructure available through the use of remote testing facilities.
"The AI Driving Olympics offer a glimpse of the challenges of creating self-driving cars,” said Emilio Frazzoli, CTO and Chief Scientist, nuTonomy (an Aptiv company) and Senior Advisor at Duckietown. “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."
The competition will use the Duckietown platform, a miniature self-driving car platform used for autonomy education and research. The Duckietown project originated at MIT in 2016 and is now used by many institutions worldwide. The competition comprises four key challenges of increasing complexity, including road following on an empty road, road following with obstacles, point-to-point navigation in a city network with other vehicles, and fleet planning for a full autonomous mobility on demand system.
Competitors will have access to a suite of professional development tools (simulators, logs, baseline implementations). Real environments called “Robotariums” will be remotely accessible for evaluation. The highest scoring entries in the robotariums will be run during the live event at NIPS 2018 to determine the overall winners. Competitors can also build or acquire their own testing facility (Duckiebots and a Duckietowns), through either open-source DIY instructions, or as rewards obtained in the Duckietown Kickstarter campaign.
The AI Driving Olympics is presented in collaboration with 6 academic institutions: ETH Zürich (Switzerland), Université de Montréal (Canada), National Chiao Tung University (Taiwan), Toyota Technological Institute at Chicago (USA), Tsinghua University (China) and Georgia Tech (USA), as well as two industry co-organizers: nuTonomy, an Aptiv company, and Amazon Web Services (AWS).
- August 1 to October 1: Open development of the competition with periodic release of simulators, tools and baseline implementations.
- October 1 to December 1: Official opening of the first AI Driving Olympiad “Robotarium” live environments available for use.
- December 7: Finals at NIPS 2018, in Montréal.
The second edition of the AI Driving Olympics, AI-DO II, will take place in May 2019 in conjunction with the International Conference on Robotics and Automation (ICRA) 2019, also in Montréal.
To learn more about the competition, please visit: http://AI-DO.duckietown.org
About Duckietown and the Duckietown Foundation
The Duckietown Foundation is a nonprofit organization dedicated to making the world excited about the beauty, the fun, the importance, and the challenges of robotics and artificial intelligence, through learning experiences that are tangible, accessible, and inclusive. Those interested can acquire their own copy of the platform via the The Duckietown Foundation Kickstarter page here. To learn more about Duckietown, watch the backstory here: https://bit.ly/2M5Cm63 or visit: https://www.duckietown.org/
Press contact: Liam Paull firstname.lastname@example.org
Duckietown and AI-DO are developed at ETH Zürich in Prof. Emilio Frazzoli’s group, at the Institute of Dynamic Systems and Control, in the Department of Mechanical and Process Engineering (D-MAVT). Duckietown is used in AI and control systems courses at ETH Zurich.
Press contact: Dr. Andrea Censi email@example.com
Université de Montréal
Duckietown and AI-DO are developed at Université de Montréal in collaboration with the MILA by Prof. Liam Paull and Prof. Yoshua Bengio.
Press contact: Prof. Liam Paull firstname.lastname@example.org
Tsinghua University in Beijing, China is the leading education and research university in China. Tsinghua has participated in Duckietown since 2016.
Press contact: Prof. (Samuel) Qing-Shan Jia email@example.com
National Chiao Tung University
Duckietown is developed at National Chiao Tung University (NCTU), Taiwan, in Prof. Hsueh-Cheng (Nick) Wang’s group. Duckietown @ NCTU has engaged over 150 students in accredited courses, and inspired over 400 K-12 students in demonstrations and other events. NCTU hosted Duckietown summer schools in 2017 and 2018, including professors from Taiwan, Korea and Indonesia, uniting the teaching community and robotics enthusiasts. In 2018, the school includes more than 100 students from 10 universities.
Press contact: Nick Wang firstname.lastname@example.org
Toyota Technological Institute at Chicago
Duckietown is developed at the Toyota Technological Institute at Chicago (TTIC) by the group of Prof. Matthew Walter.
Press contact: Prof. Matthew Walter email@example.com
At Georgia Tech, development is handled by the group of Prof. Magnus Egerstedt.
Press contact: Sean Wilson firstname.lastname@example.org
nuTonomy, an Aptiv company, is developing the world’s premier autonomous vehicle technology to radically improve the safety, efficiency, and accessibility of transportation in cities worldwide. We are committed to
nuTonomy engineers offer their expertise in embodied AI to organize the AI Driving Olympics.
Press contact: email@example.com
Amazon Web Services (AWS)
For over 12 years, Amazon Web Services has been the world’s most comprehensive and broadly adopted cloud platform. AWS provides the cloud infrastructure and machine learning services that makes the AI-DO possible. This includes Amazon SageMaker, a fully-managed service that enables developers and data scientists to quickly and easily build, train, and deploy machine learning models at any scale.
Press contact: Nina Lindsey firstname.lastname@example.org