NDT Based Localization Method

This repository contains a method for localization within a pointcloud map using NDT (Normal Distributions Transform) algorithms. It is designed to work with ROS (Robot Operating System) environments.

Localization in a Pointcloud Map (.pcd)

This method utilizes .pcd (Point Cloud Data) files for map-based localization.

How to Use

Preparation

1. PCD Map and rosbag(.bag):

   • Ensure you have a .pcd map file ready. For this example, we use map_v1.pcd.
   • The ROSbag used here is mapping_lidar.bag.
   • Place the .pcd file in the map folder within your workspace.

2. Building in Your ROS Workspace:

   • Clone the entire repository into your own ROS workspace.

Configuration

1. Map Loader:

   • Modify the map loader configuration to point to your .pcd file, like so:

     <arg name="pcd_path" default="$(find ndt_localizer)/map/map_v1.pcd"/>

2. Point Cloud Downsampling (Pre-configured):

   • If using sparse Lidar data (e.g., VLP-16), set a smaller leaf_size in launch/points_downsample.launch to 2.0.
   • For denser Lidar point clouds (e.g., VLP-32, Hesai Pandar40P, HDL-64), use a leaf_size of 3.0.

Running the Package

After completing the above steps:

1. Start ROScore:

   roscore

2. Navigate to Your Workspace:

   cd catkin_ws
   source devel/setup.bash
   rosparam set use_sim_time true

3. Launch the Package:

   • To run only the ndt_localizer package:

     roslaunch ndt_localizer ndt_localizer.launch

   • To run both scan context and ndt_localizer:

     roslaunch ndt_localizer sc_ndt_localizer.launch