loam_velodyne

Why we included this project

Anyone building lidar-based navigation on a ground robot has probably run into LOAM, and this package is one of the ROS implementations people cite when they want the algorithm in practice. The method runs in two cooperating passes: a fast scan registration that recovers pose changes from each sweep, and a slower mapping step that refines those estimates against a feature-based map, so it produces odometry and a point-cloud map in real time without GPS or wheel odometry. It builds in a ROS Indigo workspace and runs directly on Velodyne VLP16 data, which makes it a workable base for autonomous navigation and environment reconstruction. The code is compact enough to read end to end, and since the underlying method is well documented, the package doubles as a practical tutorial on how feature extraction and scan matching fit together in lidar SLAM.