Autoware in CARLA

The carla autoware bridge is now hosted and maintained here.

This repository contains a demonstrator of an autoware agent ready to be executed with CARLA.

The carla autoware integration requires CARLA 0.9.10.1. You can download it from here

CARLA autoware agent

The autoware agent is provided as a ROS package. All the configuration can be found inside the carla-autoware-agent folder.

The easiest way to run the agent is by building and running the provided docker image.

Requirements

• Docker (19.03+)
• Nvidia docker (https://github.com/NVIDIA/nvidia-docker)

Setup

Firstly clone the carla autoware repository, where additional autoware contents are included as a submodule:

git clone --recurse-submodules https://github.com/carla-simulator/carla-autoware

Afterwards, build the image with the following command:

cd carla-autoware && ./build.sh

This will generate a carla-autoware:latest docker image.

Run the agent

1. Run a CARLA server.

./CarlaUE4.sh

2. Run the carla-autoware image:

./run.sh

This will start an interactive shell inside the container. To start the agent run the following command:

roslaunch carla_autoware_agent carla_autoware_agent.launch town:=Town01

CARLA Autoware contents

The autoware-contents repository contains additional data required to run Autoware with CARLA, including the point cloud maps, vector maps and configuration files.