LiDAR Mobility Map Generator

English | 日本語

LiDAR Mobility Map Generator is a map-generation tool for autonomous vehicles and autonomous mobile robots operating in closed areas and free spaces. It accepts either LiDAR SLAM output (a point-cloud map and the trajectory estimated by SLAM) or rosbag2 data (3D LiDAR point clouds and localization estimates), and automatically generates Vector Maps and Navigation Maps.

Version 0.11.0, dated September 5, 2026, is the project's first public release. The Vector Map workflow remains a beta feature for evaluation in controlled, closed-course environments, and the Navigation Map workflow remains alpha.

Main Features

• Vector Map (Lanelet2, beta): automatically generates Vector Maps for autonomous driving and mobile robotics from LiDAR SLAM output or rosbag2. In the Web GUI, users can add a target route, speed limits, and virtual stop lines.
• Navigation Map (alpha): automatically generates a 2D occupancy-grid map and route data. The outputs can be loaded by map and route servers, including the Nav2 Map Server and Route Server, and displayed in RViz2.

Characteristics

The tool creates Vector Maps and Navigation Maps for closed areas, free spaces, and private sites where map features such as target routes, speed limits, and stop lines do not yet exist. Using LiDAR SLAM output or rosbag2 recordings, it supports map creation for transport, patrol, delivery, and other systems that follow specific routes.

The output formats are compatible with widely used open-source software, including Autoware® and Nav2. Operation has been verified with data from representative LiDARs, including Hesai, Velodyne, and Livox MID-360.

Example Results

Vector Map Example

Input Point-Cloud Map

Vector Map Editor

Generated Lanelet2 Map

Navigation Map Example (MID-360)

Input Point-Cloud Map (MID-360)

Navigation Map Editor (MID-360)

Map Usage Examples

Using a Vector Map with Autoware

docs_videos_autoware_hesai_driving_rviz2.webm

Displaying a Navigation Map with Nav2 (MID-360)

Environment

• Ubuntu 24.04
• ROS™ 2 Jazzy
• C++17
• Supported data:
  • a point-cloud map in PLY or PCD format and a trajectory in TUM format; or
  • rosbag2 containing PointCloud2 and localization estimates.
• Verified inputs:
  • For rosbag2 input, map generation has been verified with Hesai, Velodyne, and Livox MID-360 data using localization estimates produced by gicp_gnss_odom_localizer.
  • For LiDAR SLAM output, map generation has been verified with output from GLIM.

Build

export LMMG_WS="$HOME/lmmg_ws"
mkdir -p "$LMMG_WS/src"
git clone https://github.com/KariControl/lidar_mobility_map_generator.git \
  "$LMMG_WS/src/lidar_mobility_map_generator"

source /opt/ros/jazzy/setup.bash
cd "$LMMG_WS"
rosdep install --from-paths src --ignore-src -r -y
colcon build --symlink-install --packages-select lidar_mobility_map_generator
source "$LMMG_WS/install/setup.bash"

Detailed Usage

• Map Creation and Editing Operator Manual

Notes

The generated left and right Lanelet boundaries are not surveyed road boundaries. They define a virtual travel corridor derived from an automatically generated or GUI-selected centerline and the specified vehicle dimensions. Use measured values for the vehicle or robot dimensions and LiDAR extrinsics, and review the generated map before use.

License

Apache License 2.0. See LICENSE.

Trademarks

Autoware is a trademark of the Autoware Foundation. ROS is a trademark of Open Source Robotics Foundation. See TRADEMARKS.md for the complete trademark and affiliation notice.