carla-ai

AI for Carla Simulator

Installation

• Install ROS Noetic
• Follow the official guide to install Carla
• Install Carla ROS Bridge. The following changes should be made in order to compile it on Ubuntu 20.04

  diff --git a/carla_ros_bridge/src/carla_ros_bridge/bridge.py b/carla_ros_bridge/src/carla_ros_bridge/bridge.py
  index dd7838b..0a7683d 100755
  --- a/carla_ros_bridge/src/carla_ros_bridge/bridge.py
  +++ b/carla_ros_bridge/src/carla_ros_bridge/bridge.py
  @@ -531,7 +531,7 @@ def main():
                  CarlaRosBridge.CARLA_VERSION, dist.version))
              sys.exit(1)
  
  -        if LooseVersion(carla_client.get_server_version()) < \
  +        if False and LooseVersion(carla_client.get_server_version()) < \
              LooseVersion(CarlaRosBridge.CARLA_VERSION):
              rospy.logfatal("CARLA Server version {} required. Found: {}".format(
                  CarlaRosBridge.CARLA_VERSION, carla_client.get_server_version()))
  diff --git a/carla_ros_bridge/src/carla_ros_bridge/lidar.py b/carla_ros_bridge/src/carla_ros_bridge/lidar.py
  index cd920d1..0201f88 100644
  --- a/carla_ros_bridge/src/carla_ros_bridge/lidar.py
  +++ b/carla_ros_bridge/src/carla_ros_bridge/lidar.py
  @@ -57,7 +57,7 @@ class Lidar(Sensor):
          ]
  
          lidar_data = numpy.fromstring(
  -            carla_lidar_measurement.raw_data, dtype=numpy.float32)
  +            carla_lidar_measurement.raw_data.tobytes(), dtype=numpy.float32)
          lidar_data = numpy.reshape(
              lidar_data, (int(lidar_data.shape[0] / 4), 4))
          # we take the oposite of y axis
  diff --git a/pcl_recorder/CMakeLists.txt b/pcl_recorder/CMakeLists.txt
  index 068ed8b..f6a8fc1 100644
  --- a/pcl_recorder/CMakeLists.txt
  +++ b/pcl_recorder/CMakeLists.txt
  @@ -1,7 +1,7 @@
  cmake_minimum_required(VERSION 2.8.3)
  project(pcl_recorder)
  
  -add_compile_options(-std=c++11)
  +add_compile_options(-std=c++14)
  
  find_package(catkin REQUIRED COMPONENTS pcl_conversions pcl_ros roscpp
                                          sensor_msgs roslaunch)

• Install the following libraries

  sudo apt install python-is-python3 python3-rosdep libsdl-image1.2-dev libsdl-mixer1.2-dev libsdl-ttf2.0-dev libsdl1.2-dev libsmpeg-dev python-numpy subversion libportmidi-dev ffmpeg libswscale-dev libavformat-dev libavcodec-dev

• Install libxerces-c v3.1

  git clone git@github.com:apache/xerces-c.git -v v3.1.4
  cd xerces-c
  ./reconf
  ./configure
  make
  cp src/.libs/libxerces-c-3.1.so /usr/lib/

• Create a conda environment from the environment.yaml file in this repository and activate it

  conda env create -f environment.yaml
  conda activate carla-ai

• Create a catkin workspace

  mkdir catkin_ws/src
  cd catkin_ws/src
  ln -s /path/to/ros-bridge
  ln -s /path/to/carla-ai
  cd ..
  sudo rosdep init
  rosdep update
  rosdep install --from-paths src --ignore-src -r
  catkin_make

• Define the following environment variables

  export ACADOS_HOME="/path/to/acados"
  export CARLA_HOME="/path/to/carla"
  export CARLA_CATKIN_WS_PATH="/path/to/catkin_ws"

• Source the development script

  source /path/to/carla-ai/scripts/devel.zsh

Running

1. Run Carla For headless mode execute

   DISPLAY= /opt/carla-0.9.10.1/CarlaUE4.sh -opengl

   or for normal mode execute

   /opt/carla-0.9.10.1/CarlaUE4.sh

2. Run ros-bridge

   roslaunch carla_ros_bridge carla_ros_bridge.launch town:=Town06

3. Run the Carla AI module

   roslaunch carla_ai carla_ai.launch

4. Run waypoint-publisher

   roslaunch carla_waypoint_publisher carla_waypoint_publisher.launch