ORB-SLAM2-with-saved-point-cloud

This repository makes .ply file (Point Cloud Map) with ORB-SLAM2 and PointCloudLibrary.
When you execute ORB-SLAM2 with this repository, you could get point cloud map result.

Requirement

The test was conducted Ubuntu 18.04 and CUDA 10.2 with the specifications shown in the table below.

Library or Pakage	Version
OpenCV	3.4.0
Eigen	3.1.1
PCL
CMake	3.10.2
Boost	1.65

These reposiory needs same requirement with ORB-SLAM2.

How to Build

There are many subfolders to build.
In /Thirdparty folder, there are /g2o and /DBoW2.

You should build these folders first.
When you clone this repository,
git clone https://github.com/Taeyoung96/ORB-SLAM2-with-saved-point-cloud.git

You move to path ~ ORBSLAM2_with_save_pointcloud/Thirdparty.
Build 2 subfolders.

cd g2o  
mkdir build
cd build
cmake ..
make
make install

and

cd DBoW2
mkdir build
cd build
cmake ..
make
make install

Go back to path ~ ORBSLAM2_with_save_pointcloud/

You should build libuvc.

cd libuvc
mkdir build
cd build
cmake ..
make && sudo make install

cd .. and go back to path ~ ORBSLAM2_with_save_pointcloud/
Uncompress vocabulary file.

cd Vocabulary
tar -xf ORBvoc.txt.tar.gz

Then build whole ORB-SLAM2 system.

cd .. and go back to path ~ ORBSLAM2_with_save_pointcloud/

chmod +x build.sh
sh build.sh

This will create libORB_SLAM2.so at lib folder and the executables mono_tum, mono_kitti, rgbd_tum, rgbd_my, stereo_kitti, mono_euroc and stereo_euroc in Examples folder.

Prepare a Dataset