concert_localization

Localization and mapping-support launch files for the Concert robot: PCD-to-Nav2-map conversion, RTAB-Map/HDL-based global localization, AMCL-based localization, and a local EKF-based state estimator.

Launch files

pcd_to_nav2_map.launch.py

Converts a stored PCD pointcloud map into a Nav2 occupancy grid via Octomap:

PCD file -> PointCloud2 /cloud_pcd -> Octomap -> /projected_map -> Nav2 map_saver_cli

ros2 launch concert_localization pcd_to_nav2_map.launch.py \
  map_file:=/home/user/data/forest_ws/maps/nav_map.pcd \
  output_name:=nav_map \
  min_z:=0.10 \
  max_z:=1.50 \
  resolution:=0.05 \
  save_cropped_pcd:=true

Argument	Default	Description
map_file	(required)	Path to the input PCD file
output_name	PCD filename stem	Output folder name under ./maps/
min_z / max_z	-0.50 / 1.50	Height band [m] kept when building the Octomap
resolution	0.05	Octomap / occupancy grid resolution [m]
save_delay	5.0	Seconds to wait before saving, to let the Octomap accumulate
save_cropped_pcd	false	If true, also save the height-filtered Octomap pointcloud (/octomap_point_cloud_centers) as a new .pcd file
cropped_pcd_prefix	<output_dir>/<output_name>_cropped_	Prefix for the cropped PCD output, only used when save_cropped_pcd is true

Output:

./maps/<output_name>/<output_name>.yaml
./maps/<output_name>/<output_name>.pgm
./maps/<output_name>/<output_name>_cropped_<timestamp>.pcd   (if save_cropped_pcd:=true)

Note: the cropped PCD is derived from the Octomap's occupied-voxel centers, so it is voxelized at resolution and reflects occupied cells rather than a lossless crop of the original input points.

hdl_localization.launch.py

Runs HDL-based global localization against a stored PCD pointcloud map.

Argument	Description
use_sim_time	Use simulation time (useful for Gazebo)
map_file	Path to the PCD pointcloud map used for localization
auto_localize	Whether to automatically call the global localization service after a delay

rtabmap.launch.py

Lidar-only SLAM/localization pipeline built on RTAB-Map, used for testing and development of the mapping pipeline (typically against the Gazebo simulation). It chains together base odometry, lidar-scan merging, ICP scan-matching odometry, RTAB-Map itself, and a final export of the obstacle cloud to a .pcd file:

concert_odometry.launch.py  ->  base odometry (odom frame)
cloud_multi_merger.launch.py -> merges lidar scans into /merged_cloud
icp_odometry (rtabmap_odom)  -> ICP scan-matching odometry -> odom_icp frame
rtabmap (rtabmap_slam)       -> SLAM (default) or localization mode, subscribes
                                 to odom_filtered_input_scan
pointcloud_to_pcd_node       -> saves /cloud_obstacles to <pcd_file><timestamp>.pcd
                                 on shutdown

Argument	Default	Description
use_sim_time	false	Use simulation time (useful for Gazebo)
deskewing	false	Enable LiDAR deskewing in icp_odometry
rtabmap_viz	true	Intended to toggle the rtabmap_viz visualization node, but that node is currently commented out of the returned LaunchDescription — this arg has no effect right now
localization	false	If true, runs RTAB-Map in localization mode against the existing map (no new map creation, no -d database wipe); if false (default), runs full SLAM and deletes the previous map database (~/.ros/rtabmap.db) on startup via the -d argument
pcd_file	maps/pointclouds_	Prefix for the accumulated /cloud_obstacles cloud, written to <prefix><timestamp>.pcd when the node shuts down

Notes:

• localization:=false (the default) always wipes the previous RTAB-Map database — pass localization:=true to localize against an existing map without rebuilding it.
• The exported .pcd is only written on node shutdown (save_timer_sec: 0.0), so the launch must be stopped (e.g. Ctrl+C) for the file to be saved.

localization.launch.py

AMCL-based localization against a saved Nav2 map: starts nav2_map_server, nav2_amcl (configured via config/amcl_config.yaml), and the associated lifecycle manager.

Argument	Default	Description
use_sim_time	true	Use simulation time
map_file	concert_mapping/maps/robot_area.yaml	Map yaml loaded by map_server
amcl_config	config/amcl_config.yaml	AMCL parameters

local_localization.launch.py

Local state estimation: a static transform (base_footprint_ekf -> imu_link_ekf), robot_localization's EKF node (configured via config/ekf.yaml), and an IMU republisher.

Config

• config/amcl_config.yaml — AMCL parameters used by localization.launch.py
• config/ekf.yaml — EKF parameters used by local_localization.launch.py