GP-SLAM

GP-SLAM is a library implenmenting sparse Gaussian process (GP) regression for continuous-time trajectory estimation and mapping. The core library is developed by C++ language, and an optional Matlab toolbox is also provided. Examples are provided in Matlab scripts.

GP-SLAM is being developed by Jing Dong and Xinyan Yan as part of their work at Georgia Tech Robot Learning Lab.

Prerequisites

• CMake >= 2.6 (Ubuntu: sudo apt-get install cmake), compilation configuration tool.
• Boost >= 1.46 (Ubuntu: sudo apt-get install libboost-all-dev), portable C++ source libraries.
• GTSAM >= 4.0 alpha, a C++ library that implement smoothing and mapping (SAM) in robotics and vision.

Compilation & Installation

In the library folder excute:

$ mkdir build
$ cd build
$ cmake ..
$ make check  # optonal, run unit tests
$ make install

Matlab Toolbox

An optional Matlab toolbox is provided to use our library in Matlab. To enable Matlab toolbox during compilation:

$ cmake -DGPSLAM_BUILD_MATLAB_TOOLBOX:OPTION=ON -DGTSAM_TOOLBOX_INSTALL_PATH:PATH=/path/install/toolbox ..
$ make install

After you install the Matlab toolbox, don't forget to add your /path/install/toolbox to your Matlab path.

Compatibility

The GP-SLAM library is designed to be cross-platform, but it has been only tested on Ubuntu Linux for now.

Tested Compilers:

• GCC 4.8, 5.4

Tested Boost version: 1.48-1.61

Linking to External Projects

We provide easy linking to external CMake projects. Add following lines to your CMakeLists.txt

find_package(gpslam REQUIRED)
include_directories(${gpslam_INCLUDE_DIR})

Questions & Bug reporting

Please use Github issue tracker to report bugs. For other questions please contact Jing Dong.

Citing

If you use GP-SLAM in an academic context, please cite following publications:

@inproceedings{Yan17ras,  
  Author = "Xinyan Yan and Vadim Indelman and Byron Boots",
  journal = " Robotics and Autonomous Systems",
  Title = "Incremental Sparse {GP} Regression for Continuous-time Trajectory Estimation and Mapping",
  Year = {2017},
  pages="120-132",
  volume = {87}
}
@article{Dong17arxiv,
  author    = {Jing Dong and Byron Boots and Frank Dellaert},
  title     = {Sparse Gaussian Processes for Continuous-Time Trajectory Estimation on Matrix Lie Groups},
  journal   = {Arxiv},
  volume    = {abs/1705.06020},
  year      = {2017},
  url       = {http://arxiv.org/abs/1705.06020}
}

License

GP-SLAM is released under the BSD license, reproduced in the file LICENSE in this directory.