Working with PointCloud2 Data
This page describes how to interpret PointCloud2 Data from LiDAR sensors and how they can be converted to custom representations using Point Cloud Library (pcl).
Last updated
This page describes how to interpret PointCloud2 Data from LiDAR sensors and how they can be converted to custom representations using Point Cloud Library (pcl).
Last updated
PointCloud2 is a common sensor message type used for point cloud data from LiDAR sensors. I will be using Ouster 0 - 64 Channel LiDAR and Ouster 2 - 128 Channel LiDAR for my explanations. These LiDARs are 3D LiDARs.
A 64 Channel LiDAR means the sensor has 64 laser emitters stacked on top of each other. So yes, a 128 Channel LiDAR will have 128 laser emitters stacked on top of each other. These laser emitters are then rotated about the central axis.
In the case of an Ouster 64-Channel LiDAR, at one timestamp, you will have a column of laser points and a column of laser reflections received by the sensor. Each receiver sensor will give a measurement for a single "point".
With 64 "points" in this single instance, we obtain a "column" in a scan. As the emitters rotate, these "columns" are then stacked next to each other to form a 2D "picture".
In one full revolution, this 2D "picture" obtained in the end is known as a "frame" and it contains all the measurements in a single revolution.
All information about a single "frame" is then packaged into a single data structure called the PointCloud2 data structure.
Documentation on PointCloud2 Data Structure: https://docs.ros.org/en/api/sensor_msgs/html/msg/PointCloud2.html
For an Ouster 0 - 64 Channel LiDAR operating at 1024x10 Mode (1024 divisions in one revolution and 10Hz Rotation Speed)
Here is an example of a single timestamp of a LiDAR PointCloud2 Data of an Ouster 2 - 128 Channel LiDAR operating at (1024x10 Mode). Try to look at the PointCloud2 Data Documentation and see if you can decipher the data content!
