This project is currently organized using submodules. In more stable releases, it will be managed with a package manager like pip. For now in an alpha stage of development, we stay light on our feet by using submodules and poetry
NOTE: This currently only works on a Linux distribution (tested on Ubuntu 20.04 and 22.04). It also only works with Python 3.10 (to be expanded in the future).
While Carla does not require use of a GPU, it is highly recommended. Using libraries like mmdet (for perception, as in avstack-core) will require a gpu.
First, clone the repository and submodules.
git clone --recurse-submodules https://github.com/avstack-lab/carla-sandbox.git
git submodule update --recursive
Dependencies are managed with poetry. This uses the pyproject.toml file to create a poetry.lock file. To install poetry, see this page.
If docker is not installed, install with something like
curl https://get.docker.com | sh \
&& sudo systemctl --now enable docker
Verify docker with docker run hello-world.
If nvidia-docker2 is not installed, run something like
distribution=$(. /etc/os-release;echo $ID$VERSION_ID) \
&& curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg \
&& curl -s -L https://nvidia.github.io/libnvidia-container/$distribution/libnvidia-container.list | \
sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g' | \
sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list
sudo apt-get update
sudo apt-get install -y nvidia-docker2
sudo systemctl restart docker
Verify nvidia-docker2 is working with
sudo docker run --rm --gpus all nvidia/cuda:11.0.3-base-ubuntu20.04 nvidia-smi
Will take about 2-3 minutes, depending on your cpu count.
poetry install
Will take about 5 minutes the first time you download.
./initialize.sh
For example, run
cd run_carla
./run_carla_0913.sh
The examples are run as shell scripts. Therefore, you have to first activate the environment with
poetry shell
Try running a non-perception example such as:
cd examples
./run_a0_autopilot.sh
Note: You should be in the examples folder before running the example scripts.
With the carla server already running, run the dataset collection script:
poetry shell
cd examples
./run_a1_dataset_collection.sh
This will log data to the folder sim_results/. To use the dataset, we need to run postprocessing. With the poetry shell activated, navigate to submodules/lib-avstack-carla. Run the postprocessing:
python postprocess_carla_objects.py ../../examples/sim_results
This will take some time, depending on how long you ran the simulation. After postprocessing is complete, inspect the dataset by going back to the notebooks/ folder and opening test_carla_dataset.ipynb.
Docker claims permission denied
Your user must be able to run docker commands. On some configurations, this may amount to adding the user to the docker user group. Contact your system admin.
CARLA docker is not displaying
If the docker container process was previously running for a different user, it may be trying to show the display for that other user. Check if there is a process using docker ps -a to get the ID's of any containers running and use docker rm <CONTAINER ID HERE> to officially stop and remove a process before trying again.
See CONTRIBUTING.md for further details.
Copyright 2023 Spencer Hallyburton
AVstack specific code is distributed under the MIT License.