fantaosha/cpl-slam

Planar Graph-Based SLAM Using the Complex Number Representation

Note: It is important to use our provided Eigen library or CholmodSupport.h when compiling CPL-SLAM.

CPL-SLAM

CPL-SLAM is an efficient and certifiably correct algorithm that solves planar graph-based SLAM using the complex number representation.

This repository features 2 algorithms

• SE-Sync: a certifiably correct algorithm for performing synchronization over the special Euclidean group.

• CPL-SLAM: a certifiably correct algorithm for performing graph-based SLAM with and without landmarks using the complex number representation.

The performance comparison of SE-Sync and CPL-SLAM on planar graph-based SLAM is shown as follows:

Getting Started

Note: It is important to use our provided Eigen library or CholmodSupport.h when compiling CPL-SLAM.

The C++ implementation of CPL-SLAM can be built and exported as a CMake project. For a minimal working example, see examples/cpl_slam, which provides a simple command-line utility for processing .g2o files.

Acknowledgements

We use David M. Rosen's optimizaiton tools and Yixuan Qiu's Spectra in our CPL-SLAM.

Copyright and License

The C++ implementations of CPL-SLAM contained herein are copyright (C) 2018 - 2019 by Taosha Fan, and are distributed under the terms of the GNU Lesser General Public License (LGPL) version 3 (or later). Please see the LICENSE for more information.

Contact: [email protected]

cpl-slam's Issues

How to test CPL-SLAM in image sequence?

The .g2o file was provided to test in this project, and I am wondering how to test in the real image sequence like kitti?
Could you instruct me how to to configure it?