The repository implements the LOAM algorithm by Zhang et al. (2014) for realtime odometry and mapping using raw LIDAR scans, as a standalone library.
The repository is build using Bazel. It assumes local repositories for
- OpenCV
- PCL
- Eigen
Please change the locations if necessary in the WORKSPACE file.
On Ubuntu 16.04 dependencies can be fetched with:
sudo apt-get update && sudo apt-get install -y libopencv-dev libpcl-dev libeigen3-dev
To build:
bazel build //...
To run on a KITTI sequence of raw LIDAR data,
bazel build -c opt tests:loam_odom
./bazel-bin/tests/loam_odom pose_file.txt aggregated.ply $(ls $PATH_TO_KITTI_VELODYNE/*.bin | sort | xargs)
- Add some unit / functional tests.
- Remove OpenCV and PCL dependencies, and use just Eigen for portability.
- Add visualization scripts.
- Run with 2-threads (one for odometry, the other for mapping).