This repository provides the open-source plane extraction algorithm of RSS, a novel point-plane-based stereo SLAM system that fully leverages plane features to enhance accuracy and robustness. The algorithm performs real-time plane extraction using only a pair of stereo images and can be easily integrated into existing stereo-based SLAM frameworks to introduce plane constraints. For technical details and experimental results, please refer to our paper.
The code has been tested on Ubuntu 18.04 and Ubuntu 20.04.
Required libraries:
- OpenCV (tested on v3.3.0)
- Eigen3 (tested on v3.3.4)
- PCL (tested on v1.9.0)
Make sure these dependencies are properly installed before building the project.
We have validated our plane extraction algorithm on the following datasets:
You can download these datasets and provide the stereo images, timestamps, and appropriate configuration files to run the plane extraction module.
- Clone the repository:
git clone git@github.com:casia-rvg/RSS_PE.git- Configure and compile:
cd RSS_PE
mkdir build
cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
make -jIf the build completes successfully, the main executable (rss_pe) will be generated in the build directory.
Below is a brief overview of the repository structure after compilation:
RSS_PE/
├── build/ # Contains build outputs and the final executable (rss_pe)
├── CMakeLists.txt # CMake configuration file
├── examples/
│ ├── euroc_timestamps/
│ │ ├── V101.txt # Example timestamps for the EuRoC V1_01_easy sequence
│ │ └── ... # Additional timestamp files
│ └── EuRoC.yaml # Algorithm parameter file for the EuRoC dataset
├── img/ # Image resource
├── include/ # Header files for core functionalities
├── main.cpp # Main entry point of the plane extraction module
├── README.md # Project introduction and usage guide
└── src/ # Source files implementing the core functionalities
Run the plane extraction module with:
./rss_pe <left_image_folder> <right_image_folder> <timestamps_file> <config_yaml>Where:
<left_image_folder>: Path to the directory containing left camera images.<right_image_folder>: Path to the directory containing right camera images.<timestamps_file>: Path to the file listing image timestamps (e.g.,V101.txtfor the EuRoCV1_01_easysequence).<config_yaml>: Path to the YAML file with algorithm parameters (e.g.,EuRoC.yamlfor the EuRoC dataset).
Example:
./rss_pe /path/to/left/images /path/to/right/images /path/to/timestamps.txt /path/to/config.yamlIf you find this work beneficial for your research or wish to reference it in any publication, please cite it as follows:
@article{wang2024rss,
title={RSS: Robust Stereo SLAM With Novel Extraction and Full Exploitation of Plane Features},
author={Wang, Haolin and Wei, Hao and Xu, Zewen and Lv, Zeren and Zhang, Pengju and An, Ning and Tang, Fulin and Wu, Yihong},
journal={IEEE Robotics and Automation Letters},
year={2024},
publisher={IEEE}
}We sincerely hope this project benefits your research and development. If you have any questions, suggestions, or bug reports, please feel free to open an issue on GitHub.
Thank you for your interest and support!