Open source library for Velodyne 3D LIDAR point clouds processing provided by Robo@FIT group at Brno University of Technology.
Currently, the odometry estimation is main contribution of this repository. Available source code implements:
- Collar Line Segments algorithm for fast odometry estimation
- RANSAC based point cloud registration using local features from RGB image
- Visual loop detection by VFH (experimental)
Library requires following dependencies to be installed in the system:
- Eigen3 library
- OpenCV (version 2.4.9 used)
- and PCL 1.7 library.
As a simple CMake project, it can be built by executing:
cd but_velodyne_lib
mkdir bin; cd bin
cmake ..
make
Detailed description of algorithm can be found in publication (see doc/ICRA16_submission.pdf):
Velas, M. Spanel, M. Herout, A.: Collar Line Segments for Fast Odometry Estimation from Velodyne Point Clouds, ICRA 2016
For demonstration, script scripts/demo.sh has been prepared. It requires internet connection (for obtaining data samples) and successfully built library. Demo aligns 5 data frames of Velodyne LiDAR and displays resulting 3D map.