"# ws24-lidar-camera-fusion"

First Step is to convert rgb8 image to bgr8 by simultaneously running the bagfile and imageconversion.py file then run this command to save the images rosrun image_view extract_images image:=/dji_osdk_ros/main_camera_images_bgr8 filename_format:=frame_bgr%04d.jpg

as the map doesn’t exist, ensure the static transform command is running:

rosrun tf2_ros static_transform_publisher 0 0 0 0 0 0 map velodyne

We need to check if the data points are synchronized as

         /velodyne_points                                               372 msgs    : sensor_msgs/PointCloud2
         /dji_osdk_ros/main_camera_images                              1122 msgs    : sensor_msgs/Image

so we run this command

rostopic hz /velodyne_points /dji_osdk_ros/main_camera_images

rosbag record -O snippet_with_bgr8.bag /velodyne_points /dji_osdk_ros/main_camera_images /dji_osdk_ros/main_camera_images_bgr8 /dji_osdk_ros/imu /tf

saving the converted images to a new bag files

Conversion of the bag file

To use R3LIVE with the input ROS bag file with our bag files, the following transformations are necessary:

Camera Images:

Input Topic: /dji_osdk_ros/main_camera_images
Transformation: Convert images from RGB to BGR format.
Output Topic: /camera/image_color
Reason: R3LIVE requires images in the BGR format for texture and visual mapping.

LiDAR Point Cloud:

Input Topic: /velodyne_points
Transformation:
Strip unnecessary fields (ring and time) from the PointCloud2 messages.
Retain only the x, y, z, and intensity fields.
Adjust point_step and row_step to reflect the new point structure.
Output Topic: /livox/lidar
Reason: R3LIVE requires point cloud data with a simplified structure for LiDAR-based mapping.
IMU Data:

Input Topic: /dji_osdk_ros/imu
Transformation
Retain only the header, angular_velocity, and linear_acceleration fields.
Remove unused fields like orientation and covariance values.
Output Topic: /livox/imu
Reason: R3LIVE expects clean IMU data for accurate state estimation and fusion with visual and LiDAR inputs.

Transformation required
Data Type Input Topic Transformation Output Topic
Camera Images /dji_osdk_ros/main_camera_images Convert RGB → BGR /camera/image_color
LiDAR Point Cloud /velodyne_points Remove ring and time, keep x, y, z, intensity /livox/lidar
IMU Data /dji_osdk_ros/imu Retain header, angular_velocity, linear_acceleration /livox/imu