3D LiDAR Map Builder

This package allows to build 3D LiDAR map of environment by providing bag file of .mcap format with recorded odometry and LiDAR topics.

In the recorded bag robot should move calmly and visit all parts of environment from different angles, it’s important to make several loop-closures while riding.

How to use

Execute next commands inside truck/packages/lidar_map directory.

Run executable lidar_map_executable with --help command to explore available options and arguments:

ros2 run lidar_map lidar_map_executable --help

Parameters:

1. --mcap-input: path to .mcap file with a ride (required)

2. --mcap-output: path to NON-EXISTING folder for saving map (required)

3. --mcap-log: path to NON-EXISTING folder for saving map logs (optional)

4. --json-log: path to json file for saving map logs (optional)

Run the executable lidar_map_executable with the assigned --mcap-input and --mcap-output arguments:

ros2 run lidar_map lidar_map_executable
  --mcap-input data/rides/ride_atrium_XX_YY_ZZ.mcap
  --mcap-output data/clouds/cloud_atrium_XX_YY_ZZ

Logging Options

You can now specify logging options separately:

1. .mcap file: contains poses and clouds for each iteration of pose graph

2. .json file: contains poses with ICP and odom edges on each iteration of the graph

ros2 run lidar_map lidar_map_executable
  --mcap-input data/rides/ride_atrium_XX_YY_ZZ.mcap
  --mcap-output data/clouds/cloud_atrium_XX_YY_ZZ
  --mcap-log data/logs/mcap/cloud_atrium_XX_YY_ZZ_log
  --json-log data/logs/json/cloud_atrium_XX_YY_ZZ_log.json

Debug

Calculate and visualize normals for reference_cloud:

CloudWithAttributes reference_cloud_with_attr = {
  .cloud = reference_cloud,
  .normals = builder.calculateNormalsForReferenceCloud(reference_cloud),
};

serialization::writer::MCAPWriter::writeCloudWithAttributes(
    "data/logs/reference_cloud_with_attr",        // path to save mcap file
    reference_cloud_with_attr,                    // cloud
    "/cloud",                                     // topic name
    "world",                                      // frame name
    70                                            // ratio of normals to be visualized
);

Calculate and visualize outliers weights for reading_cloud:

CloudWithAttributes reading_cloud_with_attr = {
  .cloud = reading_cloud,
  .weights = builder.calculateWeightsForReadingCloud(clouds[0], clouds[1])
};

serialization::writer::MCAPWriter::writeCloudWithAttributes(
    "data/logs/reading_cloud_with_attr",          // path to save mcap file
    reading_cloud_with_attr,                      // cloud
    "/cloud",                                     // topic name
    "world"                                       // frame
);