A fast and robust ground segmentation algorithm for 3D LiDAR point clouds, using concentric zone-based region-wise processing.
Patchwork is a ground segmentation algorithm for 3D LiDAR point clouds. It identifies which points belong to the ground surface versus obstacles, using a concentric zone-based region-wise approach with ground likelihood estimation. This is essential for autonomous robots and vehicles to understand their terrain and navigate safely.
Robotics researchers and engineers working on autonomous navigation, SLAM, and perception systems that rely on 3D LiDAR data.
Developers choose Patchwork for its combination of high speed (100 Hz), robustness across varied terrains, and state-of-the-art accuracy, validated through academic publication and benchmarks.
SOTA fast and robust ground segmentation using 3D point cloud (accepted in RA-L'21 w/ IROS'21)
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Achieves 100 Hz processing using Intel TBB for parallel computation, enabling real-time operation in autonomous systems, as highlighted in the README update.
Employs a concentric zone model and region-wise ground plane fitting to handle diverse and uneven outdoor terrains effectively, ensuring reliability across varied environments.
Includes ATAT to automatically estimate sensor height from ground points, reducing manual calibration efforts, as noted in the updates section.
Validated through publication in IEEE Robotics and Automation Letters and included in ground segmentation benchmarks, providing confidence in its accuracy.
Requires careful adjustment of parameters for different environments, as detailed in the project wiki, which can be time-consuming and error-prone.
Primarily integrated with ROS2 and ROS, making it less accessible for non-ROS systems without significant code modification or adaptation.
While there's a wiki for beginners, the documentation is research-oriented, lacking detailed guides for production deployment or troubleshooting.