Skip to content

Latest commit

 

History

History
117 lines (88 loc) · 6.32 KB

Ros2Extension.md

File metadata and controls

117 lines (88 loc) · 6.32 KB
Raw

RGL ROS2 extension

The extension introduces nodes to publish PointCloud2 messages to ROS2. RGL creates a ROS2 node named RobotecGPULidar and registers publishers based on constructed RGL nodes.

Supported ROS2 distributions:

  • Humble
  • Jazzy

Supported DDS implementations for ROS2 standalone build:

  • Eclipse Cyclone DDS
  • eProsima Fast DDS

Building

RGL ROS2 extension can be built in two flavors:

  • standalone - ROS2 installation is not required on the target machine. RGL build will include all required ROS2 dependencies.
  • overlay - Assumes the existence of supported ROS2 installation on the target machine. RGL will use ROS2 libraries already installed on the system.

Before building RGL ROS2 extension, it is necessary to get the required dependencies. For some, the process has been automated - run setup.py --install-ros2-deps to get them.

Ubuntu 22/24

Prerequisites

  • Requirements listed in the main README.
  • ROS2 installed on the system and sourced:
    • ROS2 Humble for Ubuntu 22
    • ROS2 Jazzy for Ubuntu 24
  • Radar messages installed: 
    apt install -y ros-${ROS_DISTRO}-radar-msgs
  • For standalone build:
    • ROS2 cyclonedds and fastrtps packages.
    • patchelf tool. 
      apt install -y ros-${ROS_DISTRO}-cyclonedds ros-${ROS_DISTRO}-rmw-cyclonedds-cpp
apt install -y ros-${ROS_DISTRO}-fastrtps ros-${ROS_DISTRO}-rmw-fastrtps-cpp
apt install patchelf

Steps for overlay

  1. Use script setup.py with option --with-ros2.

Steps for standalone

  1. Use script setup.py with option --with-ros2-standalone
    • You can specify run-time search path(s) of RGL dependencies by adding the option --lib-rpath <paths>. This can be useful if you want RGL to search for ROS2 libraries in a different directory than RGL, e.g., if you work with a library that also provides its ROS2 libraries, such as ROS2ForUnity.
  2. Copy all ROS2 libraries from <build-dir>/lib/ros2_standalone/ to libRobotecGPULidar.so location (or location defined with --lib-rpath).

Tips for integrating RGL with AWSIM

AWSIM project includes two ROS-based plugins: ROS2ForUnity and RGLUnityPlugin. Since AWSIM is configured to run in a standalone mode (where ROS does not need to be installed on the host machine), all necessary ROS libraries are included within the project. Ros2ForUnity and RGLUnityPlugin share a common set of ROS libraries (e.g., librcl.so). To avoid duplication - since Unity does not support having multiple instances of the same libraries - one of the solutions is to store all ROS libraries in a single plugin. In this case, the ROS libraries are stored in the ROS2ForUnity plugin, and RGL is configured with a runtime search path to access them.

  1. Build RGL with the command: 
    ./setup.py --with-ros2-standalone --lib-rpath \$ORIGIN/../../../../Ros2ForUnity/Plugins/Linux/x86_64/
    $ORIGIN represents the directory in which an object (library) originated. It is resolved at run-time.
    Your rpath may differ. It depends on your relative location between RGLUnityPlugin and Ros2ForUnity.
  2. Copy library <build-dir>/lib/libRobotecGPULidar.so to the appropriate directory in your RGLUnityPlugin.
  3. Copy all ROS2 libraries from <build-dir>/lib/ros2_standalone/ to Ros2ForUnity\Plugins\Linux\x86_64 directory. Skip for duplicates.

Windows

Prerequisites

  • Requirements listed in the main README.
  • ROS2 installed on the system and sourced. We recommend installation from a pre-built binary package:
  • [ROS2 Humble only] Fixed ROS2 logging macros in rclcpp package to make it compile with C++20. More about this bug: github PR.
    • Use fix_ros2_humble.py script to apply those changes: 
      python ros2_standalone\fix_ros2_humble.py <your-path-to-ros2>

Steps for overlay

  1. Run x64 Native Tools Command Prompt for VS 2019 and navigate to RGL repository.
  2. Run python setup.py --with-ros2 command to build RGL with ROS2 extension.

Steps for standalone

  1. Run x64 Native Tools Command Prompt for VS 2019 and navigate to RGL repository.
  2. Run python setup.py --with-ros2-standalone command to build RGL with ROS2 extension and install ROS2 libraries.
  3. Copy all ROS2 libraries from <build-dir>/lib/ros2_standalone/ into RobotecGPULidar.dll location, or extend environment variable Path appropriately.

Tips for integrating RGL with AWSIM

See description from Ubuntu section for more details.

  1. Build RGL with ROS2 standalone as described above.
  2. Copy <build-dir>/RobotecGPULidar.dll and all dependent libraries located in <build-dir> to the appropriate directory in your RGLUnityPlugin.
  3. Copy all ROS2 libraries from <build-dir>/lib/ros2_standalone/ to Ros2ForUnity\Plugins\Windows\x86_64 directory. Skip for duplicates.

Note: On Windows there is no -rpath linker flag to modify an executable’s search path. Instead, the Path environment variable can be set. ROS2ForUnity handles this automatically before the simulation starts.

Usage

Each RGL node for ROS2 publishing must be connected to a format node that defines fields and their layout in the binary data. For example, to publish PointCloud2 message with fields XYZ and DISTANCE, the code should look as follow:

...
rgl_node_t nodeFormat = nullptr, nodeRos2 = nullptr;
std::vector<rgl_field_t> fieldsToPublish = { RGL_FIELD_XYZ_VEC3_F32, RGL_FIELD_DISTANCE_F32 };
rgl_node_points_format(&nodeFormat, fieldsToPublish.data(), fieldsToPublish.size())
rgl_node_points_ros2_publish(&nodeRos2, "example_topic", "example_frame");

rgl_graph_node_add_child(..., nodeFormat);
rgl_graph_node_add_child(nodeFormat, nodeRos2);

API documentation

More details can be found here.