LEO DRIVE ASSIGNMENT

Melikenur Kaymak

This repo contains the Leo Drive Internship Assingment for localization sub-team.

General Information

• Written with ROS2 and C++.
• Package contains 2 ROS2 nodes.
  • These are point_cloud_publisher and kml_parser_node.
• Package has two different folders.
  • These are src and data.
  • data has 2 .pcd files and 1 .kml file.

Dependencies

• C++
• PCL 1.2
• ROS2
• GeographicLib

Usage

After creating the package in a ROS2 workspace, which node you want to run, below are the necessary commands for both nodes

ros2 run leo_drive_assignment point_cloud_publisher

ros2 run leo_drive_assignment kml_parser_node

point_cloud_publisher

• The point_cloud_publisher node is written for registering the given point clouds into the one combined point cloud under the same frame.
• For matching the right points to the right points, aligning point clouds, combining aligned clouds, the Generalized Iterative Closest Point algorithm is used.
  • To apply the algorithm, Point Cloud Library(PCL) is used.
• All given .pcd files are written into a PCL Cloud and algorithm are run on this clouds.
  • Transform matrix between 2 given point clouds is found.
• All given and generated PCL Clouds generated by algorithm is converted into a ROS2 Cloud and visualized on Rviz2.
  • Visualisation, transform matrix can be seen below.
    

Combined Point Cloud

Terminal Output

kml_parser node

• The kml_parser_node node is written for parsing and extracting the path via coordinates in the .kml files given.
  • Coordinates in the file were assumed formatted as WGS84. In fact this has no matter, because all coordinates is converted into a local coordinate system via GeographicLib.
• After translation to the x, y, z coordinates, a local coordinate system is implemented which has the origin point as the first point in the line string coordinate array in the .kml file.
• After finding the local x, y, z coordinates, data is ready for the visualition. Path is published on ROS2 and visualised in Rviz2.
  • These results can be seen below.

Path Visualiation on Rviz2