JointLoc

[IROS 2024] JointLoc: A Real-time Visual Localization Framework for Planetary UAVs Based on Joint Relative and Absolute Pose Estimation

Update

2024.08.22

The installation and usage instructions are uploaded. You can refer to the setup.md to install and run the JointLoc.

2024.07.31

All the codes are uploaded. We are now preparing the installation and usage instructions.

2024.06.30

The paper has been accepted by IROS 2024. The the test dataset is uploaded. You can download the test dataset from Google Drive and extract it to the JointLoc directory.

We are preparing the main files (vslam.cc in RelLoc and estimate.py in AbsLoc) and will upload them soon.

2024.04.26

Most codes of the AbsLoc and RelLoc modules are uploaded. Meanwhile, the main files are deleted due to the privacy of the project. We will provide the main files and the test datset after the paper is accepted.

Getting Started

Installation

git clone https://github.com/LuoXubo/JointLoc.git
cd JointLoc

Setup the environment

You can refer to the setup.md to install the required packages.

Quick Demo

Download the dataset

Download the test dataset from Google Drive and extract it to the JointLoc directory. The given dataset and your own dataset should have the following structure:

JointLoc
├── dataset
│   ├── [Terrain]_dataset
│   │   ├── uav
│   │   │   ├── images
│   │   │   │   ├── 000000.png
│   │   │   │   ├── 000001.png
│   │   │   │   ├── ...
│   │   │   ├── images.txt
│   │   │   ├── gt.txt
│   │   │   ├── ...
│   │   ├── satelite
│   │   │   ├── 0_x0_y0.png
│   │   │   ├── 0_x0_y1.png
│   │   │   ├── ...
│   │   ├── satelite.png
│   ├── ...

Run the demo

cd JointLoc
sh RelLoc/redis.sh      # run redis server
sh AbsLoc/server.sh     # publish images to redis
sh RelLoc/relloc.sh     # run relative localization
sh AbsLoc/absloc.sh     # run absolute localization

This script will open 4 temrminals:

1. The first terminal will open redis-server.
2. The second terminal will publish the images to the redis-server.
3. The third terminal runs the relative localization (SLAM) system in /RelLoc/ directory.
4. The fourth terminal runs the absolute localization system in /AbsLoc/ directory.

Evaluation

The estimated 6-DoF poses of JointLoc will be saved in results/poses.txt. You can evaluate the performance by running the following script:

evo_ape tum dataset/groundtruth.txt results/poses.txt -pv

Compared with ORB-SLAM2 and ORB-SLAM3

We also provide the code to run ORB-SLAM2 and ORB-SLAM3 on the same dataset. You can run the following script:

sh run_orb_slam2.sh
sh run_orb_slam3.sh

Visualization

<img src="figs/comp.gif" alt="Comparison">

Acknowledgement

This work incorporates many open-source codes. We extend our gratitude to the authors of the software.

• ORB-SLAM2
• ORB-SLAM3
• SuperPoint
• LightGlue

Citation

If you find this repo useful for your research, please consider citing the paper

@misc{luo2024jointloc,
      title={JointLoc: A Real-time Visual Localization Framework for Planetary UAVs Based on Joint Relative and Absolute Pose Estimation},
      author={Xubo Luo and Xue Wan and Yixing Gao and Yaolin Tian and Wei Zhang and Leizheng Shu},
      year={2024},
      eprint={2405.07429},
      archivePrefix={arXiv},
      primaryClass={cs.RO}
}

Star History