Welcome to KITcar’s Simulation

This is the documentation of kitcar-gazebo-simulation; a ROS and Gazebo based simulation tool that can generate CaroloCup roads and simulate a car driving on the generated roads.

Note

There’s currently no dynamics simulation. The position or speed of the car must be calculated externally.

View of the Gazebo user interface displaying the default_road.

ROS topics allow to easily access sensor data and modify the car’s position or speed. Thus, allowing to completely simulate the behavior of the car.

Front Camera Output of the Car.

Additionally, we provide a generative adversarial network that is trained to modify the simulated camera image to look more realistic.

Front Camera Output of the Car with the Neural Network turned on.

Public parts of the source code are available on Github. The project was originally inspired by the drive_sim_road_generation created by TUM-Phoenix but significantly diverged since and grew into a completely independent project developed by KITcar.

This documentation aims to enable an easy entry to using the simulation while still providing a “behind-the-scenes” explanation of the implementation. First, there are several introductory pages, before diving into the different ROS packages and python packages used in the repository afterward. The documentation concludes with a list of “mini”-talks that were held internally at KITcar to share knowledge and ensure common standards within the code.

Docker Images