Duckietown is an open research project in the autonomous robotics domain developed by Duckietown Foundation. It consists of two parts – robots (“Duckiebots”) and a town ("Duckietown") where these robots roam. A lifelike traffic environment including traffic lights, road signs, road markings, and obstacles are simulated throughout the town. Duckiebots are built on an autonomous chassis and work using a single-board computer with ROS. All of the software is autonomous. The main sensor is used for receiving data from the environment.
Our laboratory uses it as a platform for developing and debugging our own solutions, some of which are further integrated into the main project:
- Autolab Remote Service
A solution that allows full remote access to the lab. Control the robots remotely, download and run your own solutions on them, and collect debugging data.
- Map Editor
Application for the visual design of a virtual environment for testing solutions in a simulator. Integrated into the Duckietown project.
- Duckietown Dashboard extensions
Extensions for the standard dashboard that allow you to visualize ROS topics or change the solution code directly in the browser when working with the robot.
- Extended localization system
Ground truth system that uses video from a small number of high-resolution surveillance cameras as input data. This approach allows a small number of cameras and markers to be used to mark the polygon while still allowing the robots' trajectories to be tracked over the entire area of the polygon.
- Auto charging system
To support the continuous operation of the robots, we are developing a way for them to charge autonomously. It includes both the docking station hardware and an algorithm for docking, determining the charge level, planning the trajectory to the charging station, and returning the robot to the experimental zone upon completion.