lidar_undistortion

This package contains libraries and ROS nodes that perform the following operations:

• Motion compensation of point clouds from a sequence of poses from an external source
• Conversion of point clouds into spherical coordinates (range, azimuth and altitude) and viceversa
• Conversion of range data into regular single channel images, for visualization purposes
• Pose buffering with interpolation

History

The code in this repo was originally developed and open sourced by the ASL group of ETH Zurich in this repo in 2020. The original developer was Victor Reijgwart.

Between 2020-2023 members of Oxford Robotics Institute added extra features to support the Hesai lidar and to fix bugs. In 2025 ORI ported the code to ROS2 and open sourced this particular repo.

The original repo applied a BSD-3 license. This has been retained for this version.

Supported devices

• Hesai XT32 and QT64
• Ouster OS1
• Velodyne VLP-16 (range conversion partially working)

ROS nodes

ouster_undistortion_node

This node subscribes to lidar point clouds in lidar frame and sensor poses in a fixed frame as sensor_msgs::PointCloud2 and geometry_msgs::PoseWithCovarianceStamped messages, respectively. Note that the covariance of the message is not used at the moment. The node outputs a motion compensated point cloud in lidar frame. Both input and output clouds are wrapped around the point definition by Ouster, which includes the following fields: x, y, z, intensity, t, reflectivity, ring, noise, range. See the official manual for more information

ouster_image_converter_node

This node subscribes to an Ouster point cloud as sensor_msgs::PointCloud2 and converts them into range, altitude and azimuth values (a.k.a. spherical coordinates). These values are published on ROS as a custom message of type lidar_undistortion::RangeImage. It also converts the spherical coordinates, intensity, and reflectivity values into single channel images and publish them.

velodyne_undistortion_node

Equivalent node for the Velodyne VLP-16 (a.k.a. Velodyne Puck). It works but has not been tested as extensively as the Ouster counterpart.

velodyne_image_converter_node

Equivalent node for the Velodyne VLP-16 (a.k.a. Velodyne Puck). It runs but the image conversion is still work in progress (the image is upside down).

Dependencies

• ouster_ros: can be cloned from here
• velodyne_pointcloud: can be installed from apt via sudo apt install ros-melodic-velodyne-pointcloud

Both dependencies are optional. However, if neither of these are found on the system, only a library with minimal capabilities is built.

Target System

This code targets ROS2 Humble running on Ubuntu 22.04.

Previous version of this code were tested on difference versions of ROS1 e.g. Melodic and Noetic.

How to Run

To run the undistortion nodes, look at the os1_undistortion.launch and vlp16_undistortion.launch launch files, edit the appropriate fields for the reference frames and topics and run them as usual.

To run the image converter node, just run it with no paramenters:

rosrun lidar_undistortion ouster_image_converter_node