How do I use lidarslam_ros2 with my own LiDAR data bag?

Run the beginner script with your rosbag2 file, ensuring it contains required LiDAR PointCloud2 and IMU topics as per the launch files, and follow the quickstart guide for configuration. The scripts like run_autoware_map_beginner.sh handle arbitrary rosbag2 inputs.

lidarslam_ros2 vs LIO-SAM: which is better for Autoware?

lidarslam_ros2 is better if you need non-GPL licensing and focused Autoware map compatibility, as it avoids GPL dependencies and includes benchmarking tools. LIO-SAM is GPL-licensed and may not have the same map-authoring workflow or verification integration.

How to filter dynamic objects in the saved pointcloud map?

Enable use_dynamic_object_filter: true in the configuration, tune parameters like voxel size and observation window, then call the /map_save service. The README shows it reduces points by 50% while keeping verification PASS.

What datasets are supported for benchmarking SLAM performance?

It includes tracked benchmarks for NTU VIRAL and MID360 datasets, with scripts to download and run them, and automated reports detailing APE RMSE and map verification results.

Does lidarslam_ros2 support real-time mapping or only offline processing?

It can run in real-time with live sensor data, but the emphasis is on map authoring and benchmarking, so performance might be tuned for accuracy over minimal latency compared to pure odometry systems.

How to add GNSS data to georeference the maps?

Enable GNSS by setting use_gnss:=true and providing NavSatFix messages on the default topic, with covariance inspection tools to ensure quality. It writes map_projector_info.yaml with georeferenced origins.

Open-Awesome

lidarslam_ros2

BSD-2-ClauseHTML0.1.0-jazzy

ROS 2 LiDAR SLAM for creating non-GPL pointcloud maps compatible with Autoware, featuring loop closure and benchmarking.

Visit Website GitHub

805 stars161 forks0 contributors

What is lidarslam_ros2?

lidarslam_ros2 is a ROS 2 package for LiDAR-based Simultaneous Localization and Mapping (SLAM) that specializes in creating and benchmarking pointcloud maps. It solves the problem of generating accurate, loop-closed maps that are directly compatible with Autoware's mapping workflows, while maintaining a non-GPL licensing default.

Target Audience

ROS 2 developers and researchers working on autonomous vehicles or robotics who need to create, benchmark, and deploy pointcloud maps for Autoware-compatible systems.

Value Proposition

Developers choose lidarslam_ros2 for its focused integration with Autoware, non-GPL licensing, comprehensive benchmarking tools, and features like dynamic-object filtering and GNSS georeferencing that enhance map quality for real-world deployment.

Overview

ROS 2 LiDAR SLAM for pointcloud-map authoring, benchmarking, and Autoware-compatible map workflows.

Use Cases

Best For

Creating Autoware-compatible pointcloud maps from LiDAR data
Benchmarking LiDAR SLAM performance on standard datasets like NTU VIRAL
Generating non-GPL maps for commercial or proprietary robotics projects
Integrating GNSS data to georeference SLAM-generated maps
Cleaning pointcloud maps by filtering out dynamic objects like vehicles and pedestrians
Conducting reproducible SLAM experiments with automated report generation

Not Ideal For

Projects needing lanelet2 map generation or full Autoware planning/localization stack integration
Teams that require GPL-licensed SLAM components or prefer established GPL stacks like LIO-SAM
Applications focused solely on real-time odometry without pointcloud map saving or benchmarking needs
Simple 2D SLAM tasks or environments with minimal sensor setups lacking LiDAR and IMU data

Pros & Cons

Pros

Non-GPL Licensing

Default workflow uses BSD-2-Clause and MIT-licensed components like graph_based_slam and RKO-LIO, avoiding GPL dependencies for commercial or proprietary projects.

Autoware Map Compatibility

Generates pointcloud maps and map_projector_info.yaml files that Autoware can load and verify, as demonstrated in the quickstart with PASS verification results.

Dynamic Object Filtering

Save-time filtering removes likely dynamic objects (e.g., cutting saved points by 50% on Leo Drive bag6) while maintaining map verification PASS, improving map cleanliness without affecting live odometry.

Comprehensive Benchmarking Suite

Includes tracked benchmarks on datasets like NTU VIRAL and MID360 with automated report generation, ensuring reproducibility and performance validation for map authoring.

Cons

Complex Setup and Dependencies

Requires cloning with submodules, ROS 2 installation, and specific dependency management via rosdep, which can be cumbersome for quick prototyping or inexperienced users.

Niche Focus and Limited Scope

Primarily designed for Autoware-compatible map authoring, so it lacks features like lanelet2 map generation and has no wheel-speed input in the current public path, limiting broader SLAM applications.

Experimental and Alpha Status

Marked as v2 alpha with features like Applanix IMU deskew being experimental, indicating potential instability or breaking changes in future updates.

Frequently Asked Questions

Related Projects

navigation

ROS 2 Navigation Framework and System

Stars4,212

Forks1,814

Last commit3 days ago

slam_toolbox

Slam Toolbox for lifelong mapping and localization in potentially massive maps with ROS

Stars2,449

Forks680

Last commit5 days ago

darknet_ros

YOLO ROS: Real-Time Object Detection for ROS

Stars2,429

Forks1,204

Last commit1 year ago

rosbridge_suite

Server Implementations of the rosbridge v2 Protocol

Stars1,195

Forks581

Last commit12 days ago

Community-curated · Updated weekly · 100% open source

Found a gem we're missing?

Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.

Submit a project Star on GitHub