Good line cutting is an enhancement module that is designed for line-assisted VSLAM, such as PL-SLAM. The primary objective of good line cutting is to pre-condition line-assisted pose optimization, therefore improving the robustness of optimization under 3D line triangulation error.
This repo is an integration of good line cutting to stereo PL-SLAM. When tested on multiple challenging scenarios that point feature may fail, GF-PL-SLAM has better performance than PL-SLAM baseline and other state-of-the-art stereo SLAM systems.
To build GF-PL-SLAM, first clone the repo to your ros workspace
git clone [email protected]:YipuZhao/GF_PL_SLAM.git
Build dependencies for GF-PL-SLAM with
./build_dep.sh
Build the GF-PL-SLAM itself
./build.sh
With the head version of g2o, it's possible to prompt error when building GF-PL-SLAM. When getting error like "no matching function for BlockSolver XXX", refer to the fix at
https://github.com/rubengooj/pl-slam/issues/29#issuecomment-437773067
To run GF-PL-SLAM, please refer to some example batch evaluation scripts at folder
batch_script
If you use GF-PL-SLAM in an academic work, please cite:
@inproceedings{zhao2018good,
title={Good Line Cutting: towards Accurate Pose Tracking of Line-assisted VO/VSLAM},
author={Zhao, Yipu and Vela, Patricio A},
booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
pages={516--531},
year={2018}
}
Notice that this repository is only an open-source version of PL-SLAM released with the aim of being useful for the community, however, it is far from being optimized and we are not including some features of PL-SLAM.
This code contains an algorithm to compute stereo visual SLAM by using both point and line segment features.
Authors: Ruben Gomez-Ojeda, David Zuñiga-Noël, Francisco Angel Moreno, Davide Scaramuzza, and Javier Gonzalez-Jimenez
Related publication: PL-SLAM: a Stereo SLAM System through the Combination of Points and Line Segments
If you use PL-SLAM in your research work, please cite:
@article{gomez2017pl,
title = {{PL-SLAM: a Stereo SLAM System through the Combination of Points and Line Segments}},
author = {Gomez-Ojeda, Ruben and Zuñiga-Noël, David and Moreno, Francisco-Angel and Scaramuzza, Davide and Gonzalez-Jimenez, Javier},
journal = {arXiv preprint arXiv:1705.09479},
year = {2017}
}
The pdf file can be found at https://arxiv.org/abs/1705.09479.
Related publications:
License:
The provided code is published under the General Public License Version 3 (GPL v3). More information can be found in the "LICENSE" also included in the repository.
Please do not hesitate to contact the authors if you have any further questions.
It can be easily found at http://opencv.org.
Installation on Ubuntu:
sudo apt-get install libboost-dev
It can be found at:
https://github.com/RainerKuemmerle/g2o.git
Installation on Ubuntu:
sudo apt-get install libyaml-cpp-dev
It can be found at:
https://github.com/rubengooj/stvo-pl
In case of using the provided representation class. Download and install instructions can be found at: http://www.mrpt.org/
If working with the most recent versions of the MRPT library you might find some issues due to hard refactoring, for which we recommend to use this version instead (the last one we tested):
https://github.com/MRPT/mrpt/tree/0c3d605c3cbf5f2ffb8137089e43ebdae5a55de3
We have modified the line_descriptor module from the OpenCV/contrib library (both BSD) which is included in the 3rdparty folder.
Executing the file build.sh will configure and generate the line_descriptor and DBoW2 modules, uncompress the vocabulary files, and then will configure and generate the PL-SLAM library for which we generate: libplslam.so in the lib folder, and the application plslam_dataset that works with our dataset format (explained in the next section).
We employ an environment variable, ${DATASETS_DIR}, pointing the directory that contains our datasets. Each sequence from each dataset must contain in its root folder a file named dataset_params.yaml, that indicates at least the camera model and the subfolders with the left and right images. We provide dataset parameters files for several datasets and cameras with the format xxxx_params.yaml.
For running SLAM we can load the default parameters file or employ the config_xxxx.yaml files provided for every dataset.
Usage: ./plslam_dataset <dataset_name> [options] Options: -c Config file -o Offset (number of frames to skip in the dataset directory -n Number of frames to process the sequence -s Parameter to skip s-1 frames (default 1)
A full command would be:
./plslam_dataset kitti/00 -c ../config/config_kitti.yaml -o 100 -s 2 -n 1000
where we are processing the sequence 00 from the KITTI dataset (in our dataset folders) with the custom config file, with an offset -c allowing to skip the first 100 images, a parameter -s to consider only one every 2 images, and a parameter -n to only consider 1000 input pairs.