Welcome to online-gpslam

[Gaussian process|incremental|GTSAM 3.2|SLAM]

Here is the code repository for the paper Incremental Sparse GP Regression for Continuous-time Trajectory Estimation and Mapping By Xinyan Yan, Vadim Indelman, and Byron Boots

Introduction

The code is dependent on GTSAM 3.2, which implements smoothing and mapping (SAM) in robotics and vision, using factor graphs and Bayes networks as the underlying computing paradigm rather than sparse matrices.

The code consists of two parts:

• cpp :c++ code that implements the factors used in the paper, including:
  • 2D Gaussian process prior factor (GaussianProcessPriorPose2.h)
  • BearingRange factor for LieVector (BearingRangeFactorLV.h)
  • Range factor for LieVector (RangeFactorLV.h)
  • Interpolated Range factor (InterpolatedRangeFactor.h)
  • Projected velocity factor (VFactor.h)
  • Interpolated velocity factor (InterpolatedVFactor.h)
• matlab: matlab scripts that produce the experimental results in the paper. We conducted experiments on one synthetic and two real-world datasets:
  • synthetic dataset with 1,500 poses
  • autonomous mower Plaza dataset available at here
  • victoria park vehicle dataset available at here

In particular, matlab scripts in names like XXX_periodic correspond to the experiments using the periodic batch update (PB) approach, and scrips in names like XXX_isam2 correspond to the experiments using the Bayes tree with Gaussian process (BTGP) approach.

Installation

The code relies on GTSAM 3.2. After installing GTSAM 3.2 and cloning the repository, just execute the following commands in a shell:

cd [repo_folder] 	# go to the repo folder
mkdir build      
cd build
cmake ..	  		# configure
make 				# build
sudo make install 	# install

Please reach me at voidpointer@gatech.edu if there's any problem. Thanks!