LOAM: Lidar Odometry and Mapping (Video)
LOAM is a 3D mapping method using lidar point cloud and optionally an IMU. The method performs scan matching to estimate motion of the lidar and map in realtime. The program has two threads running in parallel at different frequencies, for coarse to fine processing of the point cloud. The method now supports a back and forth spin lidar, a continuous spin lidar (mounted on an UAV), and the Velodyne VLP-16.
DEMO: Depth Enhanced Monocular Odometry (Video)
DEMO estimates motion of a monocular camera with assistance of depth information, i.e. from a lidar or an RGB-D camera, and optionally an IMU. Different from existing RGB-D visual odometry, the method can handle sparsely provided depth information by involving both features whose depth is available and unknown. Two versions of the code are available, using a camera + 3D lidar and an RGB-D camera.
• Slimmed down version for embedded systems (Code)