ROS 2 Velodyne Transform Node does not publish /velodyne_points at 10 Hz (Jazzy)

[ming0071]

🛠 System Information

• OS: Ubuntu 24.04
• ROS 2 Version: Jazzy
• Installation Method: Cloned from ros2 branch

📌 Issue Summary
When running velodyne_pointcloud::Transform on ROS 2 Jazzy, the node correctly subscribes to /velodyne_packets and receives data at 10 Hz. However, the output topic /velodyne_points is either published at a very low frequency (sometimes < 1 Hz) or not published at all — despite seeing the point cloud appear in RViz.

In contrast, the ROS 1 version (using the same lidar and parameters) consistently publishes /velodyne_points at 10 Hz.

🔍 Observations

Adjusting the min_range parameter in ROS 2 has a significant effect:

Low values (e.g., min_range: 0.1) → /velodyne_pointsrarely or never publishes.

Higher values (e.g., min_range: 2.0 or above) → publishing becomes stable.

It appears the node will only publish when valid points within the given min_range exist, suggesting there is some kind of filtering or accumulation threshold internally.

The cut_angle parameter was commented out and not affecting the issue.

📂 Configuration

VLP16-velodyne_driver_node-params.yaml:

velodyne_driver_node:
    ros__parameters:
        device_ip: 192.168.2.81
        gps_time: false
        # cut_angle: 0.0
        time_offset: 0.0
        enabled: true
        read_once: false
        read_fast: false
        repeat_delay: 0.0
        frame_id: velodyne
        model: VLP16
        rpm: 600.0
        port: 2368
        timestamp_first_packet: false

VLP16-velodyne_transform_node-params.yaml：

velodyne_transform_node:
    ros__parameters:
        calibration: VLP16db.yaml
        model: VLP16
        min_range: 2.0
        max_range: 130.0
        view_direction: 0.0
        fixed_frame: "base_link"
        target_frame: "velodyne"
        organize_cloud: false

❓ Questions

Has the publish behavior of velodyne_pointcloud::Transform changed in ROS 2 compared to ROS 1?

Is this expected behavior — that no points within min_range will cause the entire scan to be dropped?

Is there a threshold inside the cloud container logic that causes it to skip publishing when too few or no valid points remain after filtering?

Any insights would be appreciated!

[g1y5x3]

with the default DDS middleware I also observed the same behavior on humble. However, when switch the middleware to rmw_zenoh, I was able to consistently receive converted point clouds at 10Hz with VLP16-A.

You can install the middleware via

sudo apt update && sudo apt install ros-<DISTRO>-rmw-zenoh-cpp # replace <DISTRO> with the codename for the distribution, eg., rolling

and then follow their instructions to run

export RMW_IMPLEMENTATION=rmw_zenoh_cpp

and finally you can test with any of the launch files