Convex hull based bounding box fitting for 3D LiDAR Point Clouds

An ROS implementation of "An Efficient Convex Hull-Based Vehicle Pose Estimation Method for 3D LiDAR"

Features

• This method is faster comparing to the vanilla L-shape fitting algorithm
• It can achieve accurate vehicle pose estimation in most cases.

Known Issues

• In some cases, this method can not get the right vehicle pose.

Environment

• Ubuntu 18.04/20.04
• ROS melodic/noetic

Preparing the dataset

• Download the processed kitti dataset from https://drive.google.com/file/d/1sEsicSuYgImZmh_mg5JCbphG9JuTk7Cz/view?usp=sharing

How to use

# clone the repo
mkdir -p catkin_ws/src
cd catkin_ws/src
git clone https://github.com/HMX2013/kitti3d_gt_loader
git clone https://github.com/HMX2013/Covex-hull-based-fitting-3D-LiDAR-ROS

download obsdet_msgs from
"https://drive.google.com/file/d/1ztLk9Slm656CV-WJieUpBJPlz-Iw14Bk/view?usp=share_link"

download the processed KITTI dataset from
https://drive.google.com/file/d/1sEsicSuYgImZmh_mg5JCbphG9JuTk7Cz/view?usp=share_link

cd ../
catkin_make

source ./devel/setup.bash
roslaunch kitti3d run.launch
roslaunch convex_optimize_fitting run_rviz.launch

Citation

If our research has been helpful, please cite the below papers:

@article{ding2023efficient,
  title={An Efficient Convex Hull-Based Vehicle Pose Estimation Method for 3D LiDAR},
  author={Ding, Ningning},
  journal={arXiv preprint arXiv:2302.01034},
  year={2023}
}

Contribution

You are welcome contributing to the package by opening a pull-request

We are following: Google C++ Style Guide, C++ Core Guidelines, and ROS C++ Style Guide

License

MIT License