This repository contains the source code for the "Praktikum Autonomes Fahren" at the Chair of Mechatronics from the University of Augsburg. The goal of the project is to develop a self-driving car that can navigate through a simulated environment. The project is based on the CARLA simulator and uses the ROS framework for communication between the different components. In the future, the project aims to contribute to the CARLA Autonomous Driving Challenge. The current maintainers of the project are @JulianTrommer and @ll7.
The current state as of 2024-03-25 of the project can be seen in the following video:
To be able to execute and develop the project, you need a Linux system equipped with:
- NVIDIA GPU (everything >= RTX 3070 should be fine)
- A minimum of 16G of RAM - A minimum of 100G of free disk space
As the project is still in early development, these requirements are subject to change.
To run the project you have to install docker with NVIDIA GPU support,
nvidia-docker.
docker and nvidia-docker are used to run the project in a containerized environment with GPU support.
More detailed instructions about the setup can be found here.
To get an overview of the current architecture of the agent you can look at the general documentation here. The individual components are explained in the README files of their subfolders.
If you contribute to this project please read the guidelines first. They can be found here.
The research on existing projects we did can be found here.