AWV-MOS

version 2024-08-01

Introduction

• AWV-MOS is a novel visibility-based MOS method that utilizes an adaptive window considering the uncertainty of points.
• AWV-MOS-LIO is a online LiDAR Moving Object Segmentation system integrated with LiDAR inertial odometry.
• The dataset used for paper experiments can be downloaded from the NAS server.
• Currently, this package only provide codes for Online MOS operation.
• After the paper accept, codes that provide various functions, such as static map creation, will be uploaded.

Dependency

• ROS (tested with noetic).

  sudo apt install ros-noetic-desktop
  

• gtsam

  sudo add-apt-repository ppa:borglab/gtsam-release-4.0
  sudo apt install libgtsam-dev libgtsam-unstable-dev
  

• TBB

Install

Use the following commands to download and compile the package.

$ cd ~/catkin_ws/src
$ git clone http://git.aimlab.co.kr/seongjun/AWV-MOS-LIO.git
$ cd ..
$ catkin_make -DCMAKE_BUILD_TYPE=Release

Download ROS Bag data

• KITTI Raw dataset ROS Bag files used for paper experiments can be downloaded from the NAS server

Run the package

1. Setup environment variable:

$ cd ~/catkin_ws/
$ source devel/setup.bash

2. Run the launch file:

$ roslaunch awv_mos run_kitti.launch

3. Play existing bag files:

$ rosbag play {your_bag.bag}

Evaluation on KITTI Raw dataset

1. build with USE_EVALUATION_POINT_TYPE macro option

$ catkin_make -DUSE_EVALUATION_POINT_TYPE=ON

2. Recored the MOS result:

$ rosbag record /awv_mos/segmented_new_scan_all

3. Run the launch file & Play existing bag files:

$ roslaunch awv_mos run_kitti.launch
$ rosbag play {your_bag.bag}

4. Setup bag_record_path and bag_file_name from evaluation.launch:

    <arg name="bag_record_path" default="/path/to/recored_bag_file_folder/"/>
    <arg name="bag_file_name" default="bag_file_name.bag"/>

5. Run the launch file:

$ roslaunch awv_mos evaluation.launch

TODO

• Add static map construction function.
• Add tools for generating required bag files from dataset such as kitti raw, urbanloco.