Details

This node subscribes to /points_raw, /points_map and /current_pose and crops the original point cloud map with respect to the location of the ego-vehicle. This is published under /reduced_map. While /points_raw serves as a trigger for cropping the original point cloud map from /points_map, /current_pose continuously updates the ego-vehicle's pose and orientation to ensure that the cropping operations occur with respect to the latest vehicle pose. Since these operations are relatively expensive, the trigger callback (/points_raw) keeps a short buffer to make sure the data is still relevant.

Setup

1. Create a ROS workspace and place this node within the src directory.
2. Run catkin_make --pkg map_reduction if multiple packages are present within the workspace.
3. Please ensure that the package dependencies are met if running outside Docker: pcl_ros, pcl_conversions.

The code associated with this repository corresponds to the following research work. If you find it useful for research, please consider citing our work.

@inproceedings{paz20:mapping,
 address = {Las Vegas, NV},
 author = {David Paz and Hengyuan Zhang and Qinru Li and Hao Xiang and Henrik I Christensen},
 booktitle = {International Conference on Intelligent Robots and Systems (IROS)},
 month = {Oct},
 organization = {IEEE/RSJ},
 pages = { },
 title = {Probabilistic Semantic Mapping for Urban Autonomous Driving Applications},
 year = {2020}
}