ToySLAM: A ROS Package for 3D Point Cloud Registration using NDT for PolyU course AAE4011

Overview

ToySLAM is a lightweight ROS package for 3D point cloud registration using the Normal Distributions Transform (NDT) algorithm. Designed for educational purposes and experimentation, this package provides a modular pipeline to process, align, and visualize point clouds in real-time. It serves as a foundational framework for SLAM (Simultaneous Localization and Mapping) applications and can be extended for robotics or autonomous systems.

SLAM Dataset

Features

• Modular Architecture: Four standalone nodes for processing, alignment, mapping, and visualization.
• NDT-Based Registration: Efficient alignment of 3D point clouds using the NDT implementation.
• Real-Time Visualization: Live preview of registered point clouds and trajectory in RViz.
• Configurable Parameters: Tune algorithm behavior via ROS parameters for optimal performance.
• Configurable Parameters: Tune algorithm behavior via ROS parameters for optimal performance.

Nodes

1. lidar_subscriber_node.cpp: subscribe the 3D LiDAR point clouds from the rosbag file and save the point clouds to the PCD files.
2. ndt_omp_mapping_node.cpp: read the PCD file continuously and perform the ICP for the continuous PCD files. Output the 4x4 transformation matrix.
3. ndt_omp_node.cpp: similar to the File 2, but also plot the trajectory in the Rviz of ROS.
4. ndt_rosbag_mapping_node.cpp: Directly read the 3D point clouds from the rosbag file and perform the ICP continuously.

Folder tree

.
├── CMakeLists.txt
├── package.xml
└── src
    ├── lidar_subscriber_node.cpp
    ├── ndt_omp_mapping_node.cpp
    ├── ndt_omp_node.cpp
    └── ndt_rosbag_mapping_node.cpp

Dependencies

• ROS Noetic (or other ROS distributions)
• PCL 1.10+ (libpcl-dev, ros-<distro>-pcl-ros)
• Eigen3
• RViz for visualization

Installation

1. Clone this repository into your ROS workspace:

   cd ~/catkin_ws/src
   git clone https://github.com/weisongwen/ToySLAM
   catkin_make
   rosrun ToySLAM  ndt_rosbag_mapping_node /home/wws/Download/UrbanNav-HK_Whampoa-20210521_sensors.bag

NEW updates for further extension (Optional)

1. uwb_node.cpp

   • simulate the UWB ranging measurements and do the positioning
   • roslaunch ToySLAM fusion.launch

2. uwb_imu_node.cpp

   • UWB/IMU fusion via sliding window optimization
   • rosrun ToySLAM uwb_imu_node

3. uwb_imu_sim_node.cpp

   • UWB/IMU data simulation. Perform the least square estimation for UWB ranging measurements with visualization
   • roslaunch ToySLAM uwb_imu_fusion_sim.launch

4. uwb_imu_EKF_node.cpp

   • UWB/IMU fusion with EKF. The simulated IMU data is not correct, please use the dataset 2025-02-06-16-30-08.bag
   • rosrun ToySLAM uwb_imu_EKF_node
   • rosbag play 2025-02-06-16-30-08.bag