Prev-Talk | Next-Talk | All-Talks | Talks-Sorted | MOOS-DAWG'19

Talk-25: Duckiepond: An Initiative of Education and Research Environment for a Fleet of Autonomous Maritime Vehicles

J.-T. Huang, C.-C. Hsu, C.-T. Hung, A. Censi, M. Benjamin, C.-F. Chen, and H.-C. Wang, National Chiao Tung University, Taiwan. National Taiwan University, Taiwan. Massachusetts Institute of Technology, USA. The Duckietown Foundation

Duckiepond is an education and research environment that includes development software, education materials, and a fleet of autonomous surface vehicles “Duckieboat.’’  Duckiepond is built upon the Duckietown and the AI Driving Olympics (www.duckietown.org]), and the goal is to provide accessible learning experiences. Duckiepond supports commonly used middlewares (MOOS and ROS) via a MOOS bridge, and the software packages are containerized in Docker to be deployed in real robots or simulation, such as the Gazebo environment of the Maritime Virtual RobotX (https://bitbucket.org/osrf/vrx). The Duckieboat is built from low-cost 3D printed or commercially available parts, and includes a monocular camera, IMU, GPS, and onboard low-powered embedded computers (Jetson Nano and Raspberry Pi) that run machine learning (ML) tools.

The Duckiepond environment provides autonomy education materials for maritime missions, such as track and trail, navigation, and coordination among Duckieboats to avoid collisions using MOOS-IvP, learning-based, or combined methods. The Duckieboat has been operating in a man-made lake, reservoir and river environments. All software, hardware, and education materials are openly available.

Photos:

Categories:

  • USVs
  • Interoperability
  • Simulation