GSLAM

GSLAM (A General SLAM Framework and BenchMark)

1. Introduction

If you use this code for your research, please cite our paper GSLAM: A General SLAM Framework and Benchmark:

@inproceedings{gslamICCV2019,
	title={GSLAM: A General SLAM Framework and Benchmark},
	author={Yong Zhao, Shibiao Xu, Shuhui Bu, Hongkai Jiang, Pengcheng Han},
	booktitle={Proceedings of the IEEE International Conference on Computer Vision},
	year={2019},
	organization={IEEE}
}

@article{gslam2019,
	title={GSLAM: A General SLAM Framework and Benchmark},
	author={Yong Zhao, Shibiao Xu, Shuhui Bu, Hongkai Jiang, Pengcheng Han},
	journal={arXiv:1902.07995 },
	year={2019}
}

1.1. What is GSLAM?

1. For SLAM developers : Everyone can develop their own SLAM implementation based on GSLAM and publish it as a plugin with open-source or not.
2. For SLAM users : Applications are able to use different SLAM plugins with the same API without recompilation and implementations are loaded at runtime.

2. Documentation

Documentation: https://zdzhaoyong.github.io/GSLAM/

• Compile and Install
• Begin with GSLAM Core
  • Svar: A Tiny Modern C++ Header Brings Unified Interface for Different Languages
  • Messenger: A Tiny Class Implemented ROS Like Pub/Sub Messaging.
  • 3D Transformations: SO3, SE3, SIM3
  • Camera Models in GSLAM
  • GImage: Replace cv::Mat with A Tiny Header
  • SLAM Map Data Structure
  • Binary Built-in File Resource
  • PICMake: A CMake Tool to Write More Elegent CMakeList.txt
• Plugins of GSLAM
  • The Plugin Structure of GSLAM
  • Application Plugin for GSLAM
  • Dataset Plugin for GSLAM
  • qviz Plugin for GSLAM
• Develop SLAM with GSLAM
  • Porting ORBSLAM to GSLAM Application
  • Tools for Developing a SLAM
  • SLAM Plugins Implemented
• Evaluate SLAM Plugins with GSLAM
  • Evaluate a SLAM plugin
  • A Dockerized SLAM BenchMark
• Use GSLAM in Other Languages
  • Use GSLAM in Python

3. Start with GSLAM

3.1. Basic usages

Test modules with google test:

gslam tests --gtest_filter=*

Run a slam system with datasets:

gslam qviz orbslam play -dataset dataset_path

3.2. Supported Datasets

GSLAM now implemented serveral plugins for public available datasets. It is very easy to play different datasets with parameter "Dataset" setted:

…

The datasets are default to be played on realtime, and the play speed can be controled with "playspeed":

gslam qviz play -dataset /odomentry/color/00/mono.kitti -playspeed 2.

The following dataset plugins are now implemented:

Users can also implement dataset plugins by own.

3.3. Implemented SLAM plugins

4. Contacts

YongZhao: [email protected]

ShuhuiBu: [email protected]

ShibiaoXu: [email protected]

5. License

The GSLAM library is licensed under the BSD license. Note that this text refers only to the license for GSLAM itself, independent of its optional dependencies, which are separately licensed. Building GSLAM with these optional dependencies may affect the resulting GSLAM license.

…