Building Bonxai and NavMap Maps from a Recorded ROSBag

This HowTo shows how to generate both Bonxai and NavMap representations from a recorded ROSBag containing a PointCloud2 map, typically produced by a SLAM algorithm. The input point cloud must be referenced to the world or map frame.

—

Setup

Complete the installation steps in Build & Install first (any of APT, Pixi or source). This tutorial uses the Bonxai Maps Manager and NavMap Maps Manager plugins, which the core easynav package does not include:

• APT:

  sudo apt install \
    ros-<distro>-easynav-bonxai-maps-manager \
    ros-<distro>-easynav-navmap-maps-manager
  

• Pixi:

  pixi add \
    ros-<distro>-easynav-bonxai-maps-manager \
    ros-<distro>-easynav-navmap-maps-manager
  

• Source: already built if you followed Build from source (which also clones NavMap, needed by the NavMap Maps Manager).

Then source your workspace as described in Build and source the workspace. You will also need a recorded ROS bag containing a PointCloud2 map (this workspace does not ship one — use your own or one produced by GridMap Mapping with LidarSLAM and EasyNav).

—

Overview

You will:

1. Align coordinate frames (world → map if needed).

2. Play a pre-recorded ROS bag containing the /map point cloud.

3. Launch EasyNav with both BonxaiMapsManager and NavMapMapsManager to build the two maps simultaneously.

4. Visualize both in RViz2.

5. Save the resulting maps to disk.

—

1. Align the Frames (world → map)

If your ROS bag publishes a PointCloud2 in the frame world, you need to publish a static transform so that the map managers receive data in map.

In a new terminal, run:

ros2 run tf2_ros static_transform_publisher --frame-id world --child-frame-id map

Keep this terminal running for the entire session. If your point cloud is already in the map frame, skip this step.

—

2. Play the ROSBag

Next, replay the recorded ROS bag containing the point cloud map.

The example bag is about 1600 seconds long. To jump close to the end (around 1500 seconds, when the map is already dense) but still leave a few seconds to build the maps, run:

ros2 bag play rosbag_excavation_urjc_map_tf_only --clock --start-offset 1500

This will replay the point cloud topic /map and the TF tree needed by the map managers.

—

3. Launch EasyNav with Bonxai and NavMap Maps Managers

In a new terminal, launch EasyNav System with both map managers active.

ros2 run easynav_system system_main \
  --ros-args \
  --params-file /absolute/path/to/bonxai-navmap.dummy.params.yaml \
  -r /maps_manager_node/navmap/incoming_pc2_map:=/map \
  -r /maps_manager_node/bonxai/incoming_pc2_map:=/map

This setup assumes:

• The input cloud topic is /map.

• The QoS settings are standard (reliable, transient local).

• Both managers will build their respective maps automatically.

The file bonxai-navmap.dummy.params.yaml defines dummy plugins for all nodes except maps_manager_node.

Example configuration:

controller_node:
  ros__parameters:
    use_sim_time: true
    controller_types: [dummy]
    dummy:
      rt_freq: 30.0
      plugin: easynav_controller/DummyController
      cycle_time_nort: 0.01
      cycle_time_rt: 0.001

localizer_node:
  ros__parameters:
    use_sim_time: true
    localizer_types: [dummy]
    dummy:
      rt_freq: 50.0
      freq: 5.0
      reseed_freq: 0.1
      plugin: easynav_localizer/DummyLocalizer
      cycle_time_nort: 0.01
      cycle_time_rt: 0.001

maps_manager_node:
  ros__parameters:
    use_sim_time: true
    map_types: [bonxai, navmap]
    bonxai:
      freq: 10.0
      plugin: easynav_bonxai_maps_manager/BonxaiMapsManager
      package: easynav_indoor_testcase
      bonxai_path_file: maps/excavation_urjc.pcd
    navmap:
      freq: 10.0
      plugin: easynav_navmap_maps_manager/NavMapMapsManager
      package: easynav_indoor_testcase
      navmap_path_file: maps/excavation_urjc.navmap

planner_node:
  ros__parameters:
    use_sim_time: true
    planner_types: [dummy]
    dummy:
      freq: 1.0
      plugin: easynav_planner/DummyPlanner
      cycle_time_nort: 0.2
      cycle_time_rt: 0.001

sensors_node:
  ros__parameters:
    use_sim_time: true
    forget_time: 0.5

system_node:
  ros__parameters:
    use_sim_time: true
    use_real_time: false
    position_tolerance: 0.3
    angle_tolerance: 0.15

—

4. NavMap Build Parameters

Internally, when the NavMapMapsManager builds a mesh from an incoming PointCloud2 (incoming_pc2_map), it calls navmap_ros::from_pointcloud2() with a navmap_ros::BuildParams structure that controls mesh reconstruction quality:

Available parameters (current defaults, not yet exposed as ROS parameters)

Parameter	Description	Default
resolution	In-plane sampling resolution (m) for voxelization.	1.0
max_edge_len	Maximum triangle edge length (m).	2.0
max_dz	Maximum allowed vertical jump (m) between vertices.	0.25
max_slope_deg	Maximum slope (degrees) relative to the vertical axis.	30.0
neighbor_radius	Neighborhood radius (m) for triangle connectivity.	2.0
k_neighbors	Alternative to radius: number of nearest neighbors.	20
min_area	Minimum triangle area (m²) to reject degenerate faces.	1e-6
use_radius	Use radius-based vs. k-NN connectivity. (bool)	true
min_angle_deg	Minimum interior angle (degrees) to avoid slivers.	20.0
max_surfaces	Keep only the N largest connected surfaces (0 = all).	0

Note

These fields are defined by navmap_ros::BuildParams (see navmap_ros/conversions.hpp in the NavMap repository), but as of this writing the NavMapMapsManager constructs this structure with its default values when handling incoming_pc2_map — it does not currently read them from the navmap plugin’s YAML configuration. If you need different mesh-reconstruction quality, you must change these defaults in code (or check whether a newer release of easynav_navmap_maps_manager has since exposed them as ROS parameters).

—

5. Visualize in RViz2

Open RViz2 to monitor both map managers.

ros2 run rviz2 rviz2 --ros-args -p use_sim_time:=true

In RViz:

• Add a PointCloud2 display for Bonxai maps. - Select the topic published by BonxaiMapsManager (e.g., /maps_manager_node/bonxai/map). - QoS: Transient Local. - You can visualize the cloud as boxes with configurable size for clearer 3D structure.

• Add a NavMapDisplay (custom display type). - Select the topic published by NavMapMapsManager (e.g., /maps_manager_node/navmap/map). - QoS: Transient Local.

Figure 6 Bonxai map visualization in RViz.

Figure 7 NavMap mesh visualization in RViz.

—

6. Save the Maps

When both maps are visible and complete, you can store them to disk using their respective map manager services.

Save NavMap

ros2 service call /maps_manager_node/navmap/savemap std_srvs/srv/Trigger

The NavMapMapsManager always writes to the hardcoded path /tmp/map.navmap (this is unconditional in the current implementation, regardless of any navmap_path_file you configured). Rename and move it to your desired location (e.g. inside maps/).

Save Bonxai Map

ros2 service call /maps_manager_node/bonxai/savemap std_srvs/srv/Trigger

Unlike the NavMap manager, BonxaiMapsManager saves back to whatever path it used to load the map: if package/bonxai_path_file are left unset, it saves to /tmp/bonxai_map.pcd; otherwise it overwrites the package/bonxai_path_file location.

Warning

The example configuration above sets bonxai_path_file: maps/excavation_urjc.pcd, so calling savemap here will overwrite easynav_indoor_testcase/maps/excavation_urjc.pcd (the original source map) rather than writing to /tmp/bonxai_map.pcd. If you want to keep the original file intact, back it up first, or temporarily clear package/bonxai_path_file before calling savemap.

—

7. Summary

You have:

• ✅ Aligned frames between world and map

• ✅ Played a recorded ROS bag with point cloud data

• ✅ Built Bonxai and NavMap maps simultaneously

• ✅ Visualized them in RViz

• ✅ Saved both maps to disk for future use

These maps can now be used in EasyNav navigation stacks (e.g. with the Costmap or NavMap planners).

—

Next steps:

• Core Design and Architecture

• Navigating with SimpleStack and EasyNav

• Navigating with the Costmap Stack