Running norlab_icp_mapper example
The library comes with an example that builds a map from individual scans and a trajectory. Use this toy example to get familiar with the mapper's configuration before jumping to integrating the mapper with ROS.
The example is located in examples/build_map_from_scans_and_trajectory.cpp and is build with the -D BUILD_EXAMPLE=TRUE flag.
After building the library, you can execute it on the provided data:
cd build
build_map_from_scans_and_trajectory ../examples/data ../examples/config.yaml
examples/data/map.vtk.
Use ParaView or other point cloud viewer to inspect it.
The config.yaml file contains configuration for the mapper, as well as for the registration algorithm handled by libpointmatcher.
Check the MapperConfiguration tutorial for more details.
The default final map looks something like this:
Input data format
In case you want to run the mapper offline on your data, you can have a look in the examples/data folder for the required format.
Scans
The scans (point clouds readings) are located in the examples/data/scans folder and can be in any format supported by libpointmatcher.
The file name must start with cloud_, followed by a timestamp representing the seconds and nanoseconds of the data recording.
For example, from the two first scans: cloud_1690309709_285305600.vtk and cloud_1690309709_385259008.vtk you can notice that they were taken with approximately 100 ms delay, indicating a sensor operating at 10 Hz.
Trajectories
Trajectories are described in examples/data/trajectory.csv.
The file format corresponds to the definition of the ROS PoseStamped message.
Importantly, the header.stamp.sec and header.stamp.nanosec columns must match the scan file names.
No interpolation in either scans or poses is happening, the number of rows (without the header) in the .csv file must exactly match the number of files in examples/data/scans.
How to get these files from a rosbag?
The mapper currently doesn't support the processing of rosbag files. For this feature, take a look at the norlab_icp_mapper_ros bridge. However, it is possible to generate files in the required format from an existing rosbag file. In fact, that is exactly how the files in this guide were created! On a system with either ROS or ROS 2 installed, download a script that exports a ROS message type into .csv and execute it:
wget https://raw.githubusercontent.com/norlab-ulaval/norlab_scripts/main/extract_topic_to_csv.sh
extract_topic_to_csv.sh input.bag /odometry trajectory.csv
wget https://raw.githubusercontent.com/norlab-ulaval/norlab_scripts/main/extract_topic_to_csv_ros2.py
python3 extract_topic_to_csv_ros2.py bagfolder trajectory.csv /odometry
This extract the topic /odometry from a rosbag (or bag folder in ROS 2) into trajectory.csv.
To get the scans from your bagfile, you'll need to have libpointmatcher installed with its Python bindings.
Check this installation guide how to do that.
Execute the following code, which fetches the exporting script and then runs it, waiting for incoming message on topic /lidar_data.
wget https://raw.githubusercontent.com/norlab-ulaval/norlab_scripts/main/save_vtk.py
save_vtk.py /lidar_data scans
rosbag play input.bag
ros2 bag play bagfolder
Your scans will be exported into the scans folder.