场景模型>> gpslam>> 返回
项目作者： gtrll

项目描述 ：
Sparse Gaussian Processes for SLAM
高级语言： C++
项目地址: git://github.com/gtrll/gpslam.git
创建时间： 2017-08-30T02:15:38Z
项目社区：https://github.com/gtrll/gpslam
开源协议：Other
下载


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:

  1. $ mkdir build
  2. $ cd build
  3. $ cmake ..
  4. $ make check  # optonal, run unit tests
  5. $ make install

Matlab Toolbox

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

  1. $ cmake -DGPSLAM_BUILD_MATLAB_TOOLBOX:OPTION=ON -DGTSAM_TOOLBOX_INSTALL_PATH:PATH=/path/install/toolbox ..
  2. $ 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

  1. find_package(gpslam REQUIRED)
  2. 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:

  1. @inproceedings{Yan17ras,  
  2.   Author = "Xinyan Yan and Vadim Indelman and Byron Boots",
  3.   journal = " Robotics and Autonomous Systems",
  4.   Title = "Incremental Sparse {GP} Regression for Continuous-time Trajectory Estimation and Mapping",
  5.   Year = {2017},
  6.   pages="120-132",
  7.   volume = {87}
  8. }
  9. @article{Dong17arxiv,
  10.   author    = {Jing Dong and Byron Boots and Frank Dellaert},
  11.   title     = {Sparse Gaussian Processes for Continuous-Time Trajectory Estimation on Matrix Lie Groups},
  12.   journal   = {Arxiv},
  13.   volume    = {abs/1705.06020},
  14.   year      = {2017},
  15.   url       = {http://arxiv.org/abs/1705.06020}
  16. }

License

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