This repository contains source codes and demo files for our paper Online Continuous Mapping using Gaussian Process Implicit Surfaces (GPIS), which is presented at IEEE ICRA 2019.
The previous repo [GPisMAP(https://github.com/leebhoram/GPisMap.git) is not maintained but kept as legacy. This repo contains some updates including slightly enhanced speed and simple python interfaces.
Licensed under GNU General Public License version 3.
-
(Option 1) Build the source via MATLAB mex and run demo & visualization scripts on MATLAB Recommended for best visualization.
-
(Option 2) Alternatively, build the source via cmake and run demo & visualization scripts in Python (tested with version 3.7 - 3.9 and CMake 3.20+). Visualization is limited, but useful in case you have no Matlab license.
- Clone this repository
git clone https://github.com/leebhoram/GPisMap2.git
cd GPisMap2
- cd to the mex directive in MATLAB
cd <GPisMap2>/matlab/mex
-
Compile the mex functions by executing the make script.
- Setup mex
mex -setup
- Run the make scripts
make_GPisMap make_GPisMap3
-
Run the demo scripts
First,
cd <GPisMap2>/matlab
Then,
- For 2D
run('demo_gpisMap.m')
- For 3D
run('demo_gpisMap3.m')
- Trouble shooting
- If mex complains about not finding eigen, configure the eigen path appropriately
in both
make_GPisMap.m
andmake_GPisMap3.m
- If mex complains about not finding eigen, configure the eigen path appropriately
in both
- Build the source
mkdir build && cd build
cmake ..
make -j $(nproc)
- Run the demo scripts
cd <GPisMap2>/python
- For 2D
python test.py
- For 3D
python test3d.py
- Python required packages (To-do)
Bhoram Lee E-mail: <first_name>.<last_name>@gmail.com
Code has been developed and tested on Ubuntu 22.04
If you find GPisMap/GPisMap_v2 useful in your research, please consider citing:
@article{<blee-icra19>,
Author = {Bhoram Lee, Clark Zhang, Zonghao Huang, and Daniel D. Lee},
Title = {Online Continuous Mapping using Gaussian Process Implicit Surfaces},
Journal = {IEEE ICRA},
Year = {2019}
}