CARLA in Docker
Users can pull an image based on a CARLA release to run in a Docker container. This is useful for users who:
- Want to run CARLA without needing to install all dependencies
- Run multiple CARLA servers and perform GPU mapping
- Run the CARLA server without a display
This tutorial explains the requirements to run the CARLA image and how to run the image with both OpenGL and Vulkan graphics APIs.
Before you begin
You will need to have installed:
- Docker: Follow the installation instructions here.
- NVIDIA Container Toolkit: The NVIDIA Container Toolkit is a library and toolset that exposes NVIDIA graphics devices to Linux containers. It is designed specifically for Linux containers running on Linux host systems or within Linux distributions under version 2 of the Windows Subsystem for Linux. Install the
nvidia-docker2package by following the instructions here.
Docker requires sudo to run. Follow this guide to add users to the docker sudo group.
Running CARLA in a container
1. Pull the CARLA image.
You can pull either the latest CARLA image or a specific release version. The latest image refers to the most recent packaged release. To pull the image, run one of the following commands:
# Pull the latest image docker pull carlasim/carla:latest # Pull a specific version docker pull carlasim/carla:0.9.12
2. Run the CARLA container.
Different versions of CARLA support different graphics APIs which can affect the conditions in which the Docker image can run:
- 0.9.12 supports only Vulkan
- 0.9.7+ supports both Vulkan and OpenGL.
To run CARLA with a display:
sudo docker run --privileged --gpus all --net=host -e DISPLAY=$DISPLAY carlasim/carla:0.9.12 /bin/bash ./CarlaUE4.sh
To run CARLA in off-screen mode:
sudo docker run --privileged --gpus all --net=host -v /tmp/.X11-unix:/tmp/.X11-unix:rw carlasim/carla:0.9.12 /bin/bash ./CarlaUE4.sh -RenderOffScreen
CARLA 0.9.7 to 0.9.11
To run CARLA using Vulkan:
sudo docker run --privileged --gpus all --net=host -e DISPLAY=$DISPLAY -e SDL_VIDEODRIVER=x11 -v /tmp/.X11-unix:/tmp/.X11-unix:rw carlasim/carla:0.9.11 /bin/bash ./CarlaUE4.sh -vulkan <-additonal-carla-flags>
This command will allow you to run the CARLA image with Vulkan as long as your machine has a display. See the rendering documentation for information on running with Vulkan in off-screen mode.
To run CARLA using OpenGL:
docker run -e DISPLAY=$DISPLAY --net=host --gpus all --runtime=nvidia carlasim/carla:<version> /bin/bash CarlaUE4.sh -opengl <-additonal-carla-flags>
3. (Optional) Configure Docker flags.
The above commands use some Docker flags that can be configured according to your needs:
- Networking: The
--net=hostargument will allow the container to share the host's entire network. If you prefer to map specific ports on the host machine to container ports, use the flag
- GPUs: You can choose to use all GPUs with
--gpus all, or target specific GPUs with
--gpus '"device=<gpu_01>,<gpu_02>"'. See here for more information.
OpenGL requires no configuration if you are running CARLA on a machine without a display, however you will need to perform some extra steps to do the same using Vulkan prior to CARLA 0.9.12. See the rendering documentation for information.
Any issues or doubts related with this topic can be posted in the CARLA forum.