ROS2 ORB-SLAM 3

ROS2 node for ORB-SLAM 3.

Currently, supports Monocular and RGB-D modes. The node subscribes to /camera/rgb/image_color (monocular) or to /camera/rgb/image_color AND /camera/depth/image (RGB-D) and publishes:

• Camera pose (/pose_orb1)
• The map points being tracked (/tracked_mappoints)
• All map points (/all_mappoints)
• The anotated tracking image (/tracking_image)
• The keyframe ref. frames (/kf_markers)
• Odometry messages (/odom)

The implementation is based on this and this implementations.

Tested on:

• Ubuntu 24.04 LTS
• ROS2 Jazzy

Install

1. Download and compile ORB-SLAM 3.
2. Install the required ROS2 dependencies:

• ros-jazzy-image-transport
• ros-jazzy-cv-bridge
• ros-jazzy-vision-opencv
• ros-jazzy-message-filters

3. Download or clone this repo under your <ROS2_Workspace>/src folder. ( i.e: /home/user/ros2_ws/src)

4. Set the install path of your ROS2 distro in the CMakeLists.txt file:

   set(ENV{PYTHONPATH} "/opt/ros/<ROS2_site_packages>")

   In my case, the line should be set to:

   set(ENV{PYTHONPATH} "/opt/ros/jazzy/lib/python3.12/site-packages")

5. Set the root path of your ORB-SLAM3 folder in the CMakeModules/FindORB_SLAM3.cmake file:

   set(ORB_SLAM3_ROOT_DIR "<path_to_your_ORBSLAM3_root_folder>")

   In my case, the line should be set to:

   set(ORB_SLAM3_ROOT_DIR "/home/bruno/Projects/ORB_SLAM3")

6. Source your ROS2 installation:

   $ source /opt/ros/<ros2_distro>/setup.bash

7. Build the package:

   $ cd /home/user/ros2_ws

   $ colcon build

Launching

1. Source your ROS2 workspace:

   $ source /home/user/ros2_ws/install/setup.bash

2. Launch the node:

   $ ros2 run ros2_orbslam3 mono <path_to_vocabulary> <path_to_settings> (monocular mode)

   or

   $ ros2 run ros2_orbslam3 rgbd <path_to_vocabulary> <path_to_settings> (RGB-D mode)