- A “class-in-a-box” that provides a novel learning experience for many fields of autonomy, from controls to AI.
- A modular robotics ecosystem comprising fleets of self-driving cars and customizable smart cities (Duckiebots and Duckietowns), with state of the art (a) software and (b) extensive pedagogical documentation (the "Duckiebook").
- An international community of responsive university level educators, stay at home “makademics” (makers with the spirit of academics), and students.
- A cost-effective research platform for robotics and embodied AI.
- The hardware platform used in the AI Driving Olympics (AI-DO) robotics competition that will be held at NIPS 2018 and ICRA 2019.
- A worldwide robotics dissemination initiative designed to be accessible to everybody in the world. Through the Duckietown Foundation, we have a sponsorship and donations program in place to subsidize the hardware costs of the platform for low income geographical areas worldwide.
How is it novel?
- It's a one stop solution for teaching an "Introduction to self-driving vehicles" course. It includes the hardware, the software, and the material for a variety of educational experiences (slides, notes, exercises, instruction sets for reproducing demonstrations that reinforce the curriculum, and much more) on a variety of topics.
- It's about systems: differently from other educational robots, the Duckietown platform includes the robots as well as the environments robots operate in. Both are "robotic", as the smart elements of the city use the same hardware architecture as the robots.
- It's realistic: differently from other educational robots, the perception pipeline is completely computer vision based. Moreover, the platform is fully decentralized. Students appreciate the challenges of real self-driving cars: having to trade off theory and practical implementation; mitigating desired performance and constrained computational resources, both at single robot and fleet level.
How is it useful for you?
- It's accessible: through the Duckietown platform you can teach core concepts of robotics, computer vision and programming as well as marketable skills, with the ease of a click.
- It's versatile: the class-in-a-box is a great starting point to teach an Introduction to autonomous vehicles course, but the Duckietown platform is designed to be modular so that any "block" can be investigated in detail without worrying about the rest of the system, that stays "under the hood".
- It's fun: Duckietown is affordable enough for every student to have her/his own Duckiebot to take at home. Duckies can be customized as well!
- It's proven to work: over the last two year, the Duckietown curriculum has been used to teach over 700 students in 3 continents at graduate and undergraduate level.
What do we want from you?
- We invite you to join our community.
- We would like you to support our Kickstarter.
- We want you and your students to participate in the upcoming AI Driving Olympics.
- We ask you to share this announcement broadly.