Skip to content

Latest commit

 

History

History
256 lines (184 loc) · 8.34 KB

README.md

File metadata and controls

256 lines (184 loc) · 8.34 KB

Dynamic LiDAR Point Cloud Interpolation

cover

ROS2 Real-time 3D LiDAR point cloud interpolation package in C++, featuring:

  • Runtime Configuration: Adjust on the fly.
  • Optimized Performance: Fast and efficient.
  • Full Control: Precision interpolation.

Watch demo: video

demo

🛠️ Key Capabilities

  1. Dynamic Interpolation
    Generates dense, continuous LiDAR point clouds with customizable settings.

  2. Noise Reduction & Resampling
    Reduces noise and standardizes point spacing for high-quality outputs.

  3. Flexible Methods
    Offers multiple interpolation options:

    • Bilinear
    • Bilateral
    • Edge-Aware
    • Spline
    • Nearest Neighbor

  4. High Performance
    Built in C++ for real-time processing with ROS2 compatibility.

  5. Cross-Platform
    Core interpolation is ROS2 independent pure C++, PCL and Eigen implementation ensures compatibility across different OS and platforms.

  6. Customizable
    Supports real-time parameter adjustments for adaptable applications.

  7. Easy Integration
    Seamlessly integrates into robotics and sensor fusion workflows.

🌐 Key Applications

  1. Autonomous Vehicles
    Enhances SLAM, mapping, and collision avoidance with smoother LiDAR data.

  2. Robotics
    Improves environmental perception for better decision-making.

  3. Digital Twins & Simulations
    Produces high-density point clouds for realistic virtual environments.

  4. Temporal Synchronization
    Aligns LiDAR with other sensors for accurate sensor fusion.

  5. Motion Distortion Correction
    Corrects rolling shutter effects for better mapping and object detection.

  6. Sparse Cloud Enhancement
    Densifies data for improved detection and segmentation performance.

  7. Splatting Map Generation
    Smooths point clouds using Gaussian splatting for advanced simulations.

🔍 Why Interpolation Matters ?

LiDAR sensors often produce sparse or inconsistent point clouds due to environmental conditions, affecting perception in autonomous systems. This package provides advanced interpolation techniques, ensuring complete and reliable point clouds for enhanced perception.

📦 Installation

Prerequisites

Ensure your system has ROS2 installed, along with the necessary dependencies:

sudo apt update
sudo apt install ros-humble-pcl-ros
sudo apt install libeigen3-dev

Clone and Build

  1. Clone the repository:

    git clone https://github.com/geekgineer/dynamic_lidar_interpolation.git
cd dynamic_lidar_interpolation

  2. Build the workspace using colcon:

    colcon build

  3. Source the workspace:

    source install/setup.bash

🚀 Running the Package

Note

Refer to the interpolation_config file to customize the behavior

Make sure to download the sample bag from here to test the interpolation node with a recorded ROS2 bag file:

ros2 bag play data/velodyne_vlp16_outdoor --rate 1.0 --loop --clock

Then, launch the interpolation node using log level debug, info, warn, error, or fatal as needed:

colcon build --packages-select dynamic_lidar_interpolation --cmake-clean-cache

ros2 launch dynamic_lidar_interpolation pointcloud_interpolation_launch.py log_level:=info

Tip

For production environments, you might want to limit debug logs to reduce verbosity.

🧩 Ranked Interpolation Methods (From Fastest to Slowest)

Rank Method Description Computational Efficiency
1 Nearest Neighbor Assigns the value of the nearest point to each interpolated point, effective for filling sparse regions. ⭐⭐⭐⭐⭐ (Very High Efficiency)
2 Bilinear Considers four nearest neighbors for interpolation, commonly used for quick upscaling. ⭐⭐⭐⭐ (High Efficiency)
3 Bilateral Balances spatial proximity and intensity similarity, preserving edges and reducing noise. ⭐⭐ (Moderate Efficiency)
4 Edge-Aware Maintains object boundaries by preserving edges, ideal for object detection tasks. ⭐⭐ (Moderate Efficiency)
5 Spline Uses piecewise polynomials for smoother transitions, providing gradual interpolations. ⭐ (Low Efficiency)

✅ Tested For

  1. LiDAR Types

    • Velodyne: VLP-16
    • Ouster: OS0, OS1, OS2 series,OSDome..
    • Robosense: RS-BPearl.

  2. Platforms

    • Tested extensively with ROS2 Humble middleware for seamless integration.
    • Compatible across Linux-based systems, including Ubuntu 20.04 and 22.04.

  3. Visualization Tools

    • Outputs verified with tools such as RViz, Foxglove Studio for point cloud analysis.
    • Works well with third-party simulation tools like Isaac Sim, Gazebo and CARLA.

📜 License

This project is licensed under the Affero General Public License (AGPL) version 3.0.

Acknowledgment of Related Work

This project builds upon and is inspired by several foundational works in LiDAR data interpolation and augmentation. The following references have been instrumental in guiding the methodologies and approaches adopted:

  1. Velasco-Sánchez, E., de Loyola Páez-Ubieta, I., Candelas, F. A., & Puente, S. T. (2023).
    LiDAR data augmentation by interpolation on spherical range image.
    Proceedings of the 2023 IEEE 28th International Conference on Emerging Technologies and Factory Automation (ETFA), pp. 1–4. IEEE.
    Link to Paper

    @inproceedings{velasco2023lidar,
  title={LiDAR data augmentation by interpolation on spherical range image},
  author={Velasco-S{\'a}nchez, Edison and de Loyola P{\'a}ez-Ubieta, Ignacio and Candelas, Francisco A and Puente, Santiago T},
  booktitle={2023 IEEE 28th International Conference on Emerging Technologies and Factory Automation (ETFA)},
  pages={1--4},
  year={2023},
  organization={IEEE}
}

  2. Xia, O., Zhao, L., & Zhou, Q. (2022).
    PolarMix: A General Data Augmentation Technique for LiDAR Point Clouds.
    arXiv preprint arXiv:2208.00223.
    Link to Paper

    @article{xia2022polarmix,
  title={PolarMix: A General Data Augmentation Technique for LiDAR Point Clouds},
  author={Xia, O. and Zhao, L. and Zhou, Q.},
  journal={arXiv preprint arXiv:2208.00223},
  year={2022},
  url={https://arxiv.org/abs/2208.00223}
}

  3. Shepard, D. (1968). A Two-Dimensional Interpolation Function for Irregularly-Spaced Data. Proceedings of the 1968 ACM National Conference, pp. 517–524. ACM.

    @inproceedings{shepard1968twodimensional,
title={A Two-Dimensional Interpolation Function for Irregularly-Spaced Data},
author={Shepard, Donald},
booktitle={Proceedings of the 1968 ACM National Conference},
pages={517--524},
year={1968},
organization={ACM}
}

  4. Multiquadric Equations of Topography and Other Irregular Surfaces. Journal of Geophysical Research, 76(8), pp. 1905–1915.

    @article{hardy1971multiquadric,
title={Multiquadric Equations of Topography and Other Irregular Surfaces},
author={Hardy, Ronald L.},
journal={Journal of Geophysical Research},
volume={76},
number={8},
pages={1905--1915},
year={1971}
}

  5. Sibson, R. (1981). A Brief Description of Natural Neighbor Interpolation. In Interpolating Multivariate Data, pp. 21–36. John Wiley & Sons.

    @inbook{sibson1981natural,
title={A Brief Description of Natural Neighbor Interpolation},
author={Sibson, Richard},
booktitle={Interpolating Multivariate Data},
pages={21--36},
publisher={John Wiley \\& Sons},
year={1981}
}

👤 Author