This ROS2 package converts a Point Cloud to a Laser-Scan message taking into consideration relative motion of the sensor. Our algorithm allows specifying constraints in the world frame, and using the camera-robot transformation, it automatically extracts the desired filtered point cloud. This allows to consistently filter the wanted depths points even when the camera is tilted. Also, our package allows to detect holes and consider them as obstacles.
|
| More materials |
This package can be used to fuse camera data with other sensors using the package multi-laserscan-toolbox-ros2.
https://github.com/Noceo200/depth-filter-scan-converter.git
- ROS2 (Tested on Humble): https://docs.ros.org/en/humble/Installation/Ubuntu-Install-Debs.html
The following ROS2 packages are used and should be available once ROS2 is installed:
- ament_cmake
- rclcpp
- sensor_msgs
- tf2
- tf2_ros
- tf2_geometry_msgs
- rosgraph_msgs
- nav_msgs
- PCL
- pcl_conversions
If needed, install the missing ones:
sudo apt-get install ros-<ros_version>-<PACKAGE_name>
cd <your_ros2_workspace>/src/
git clone https://github.com/Noceo200/depth-filter-scan-converter.git
cd ..
colcon build
| Topic | Type | Description |
|---|---|---|
| /cloud | sensor_msgs/PointCloud2 |
The specified Point Cloud input from your sensor |
| /odom | nav_msgs/Odometry |
The odometry of your robot is optionally used to compensate robot's motion and late data |
| tf | N/A | A valid transform from your configured frames |
| Topic | Type | Description |
|---|---|---|
| /scan | sensor_msgs/LaserScan |
Output Laser-Scan |
rate - The frequency at which the node operates, in Hz. (Better if equal or inferior to the depth sensor publishing speed)
topic_in - The topic name for the input point cloud.
ref_frame - The reference frame for the 3D position of points, should be a frame attached to the robot at the floor level and parallel to it. If you use a frame above the floor level like "base_link" usually, you can adjust the offset below with height_offset.
height_offset - The height offset from the floor, distance between the floor and ref_frame. (Following z axis in ref_frame, usually >0 as ref_frame above the floor)
min_height - The minimum height of a point to be considered (represent height following z axis in ref_frame).
max_height - The maximum height for points (represent height following z axis in ref_frame).
angle_min - The minimum angle of points to keep in ref_frame (not in depth sensor frame) [-pi, pi].
angle_max - The maximum angle of points to keep in ref_frame (not in depth sensor frame) [-pi, pi] and > angle_min.
range_min - The minimum range of the scan in meters (radius from ref_frame center).
range_max - The maximum range of the scan in meters (radius from ref_frame center, needs to be smaller than the maximum range of the depth camera, otherwise infinite points of the camera might be interpreted as obstacles).
speed_up_h - An integer value that allows skipping horizontal lines of points and having a better rate, 1 = all the points, 2 = 1/2 of the points, etc.
speed_up_v - An integer value that allows skipping vertical lines of points and having a better rate, 1 = all the points, 2 = 1/2 of the points, etc.
compensate_move - A boolean flag that, if true, compensates for temporary offsets on the points during movement if the computation time is not fast enough.
odom_topic - The topic name for odometry data, used if compensate_move is true to compute how ref_frame moved from world_frame during the computation, and apply a correction.
publish_cloud - A boolean flag to choose whether to publish the filtered points cloud or not. No need to publish it for the LaserScan, but can be used as debug to check which points are considered to compute the scan.
topic_out_cloud - The topic name for the output filtered point cloud.
publish_scan - A boolean flag to choose whether to publish the LaserScan or not.
topic_out_scan - The topic name for the output LaserScan message, which will be published in the frame ref_frame.
h_angle_increment - The horizontal angle increment in radians for the scan message (e.g., 0.004363323 rad for 1-degree resolution). A smaller value allows for a scan message with more points (if the cloud allows it).
cliff_detect - A boolean flag to detect and consider holes or cliffs in the floor as obstacles.
cliff_height - The height of points that should be considered as holes and therefore as obstacles (usually <0).
time_increment - The time increment in seconds. Let it be 0.0 if unknown.
scan_time - The scan time in seconds. Let it be 0.0 if unknown.
debug - A boolean flag to enable or disable debugging.
debug_file_path - The file path where debug information will be saved.
show_ranges - A boolean flag to enable or disable the display of ranges of the output scan.
Two parameters can be specified when launching the package.
use_sim_time - If True, the package will subscribed to /clock and use this time to publish the Laser-Scan and Point Cloud messages. Otherwise, the system's clock will be used.
params_file - YAML file tu consider for the configurations.
To launch with the default configurations in "config/depth_filter_scan_converter_params.yaml":
ros2 launch depth-filter-scan-converter depth_filter_scan_converter.launch.py use_sim_time:=<true/false>
To launch with your own configurations:
ros2 launch depth-filter-scan-converter depth_filter_scan_converter.launch.py params_file:=<your_yaml_file_path> use_sim_time:=<true/false>