finds.dev

HKUST-Aerial-Robotics/A-LOAM

Advanced implementation of LOAM

View on GitHub

★ 2,449C++NOASSERTIONupdated Oct 2023

// the take

A-LOAM is a cleaned-up, Ceres Solver-based reimplementation of the classic LOAM lidar odometry/mapping algorithm, aimed at people who want to study or build on LOAM without wading through the original's gnarly math. It targets ROS1 (Kinetic/Melodic) and ships launch files for Velodyne VLP-16/HDL-32/64 and KITTI.

Replacing LOAM's hand-rolled optimization with Ceres and Eigen strips out a lot of boilerplate and makes the scan-matching cost functions (see lidarFactor.hpp) actually readable - genuinely useful if you want to understand or modify the math rather than treat it as a black box. It covers the full odometry + mapping pipeline with ready launch configs for several real lidar models plus a KITTI bag converter, so you can get point-cloud output running on real or benchmark data quickly.

It's pure LOAM front-end: no loop closure or global pose graph optimization, so long runs will drift with no built-in way to correct it - you'd need to bolt on something like SC-LeGO-LOAM yourself. It's pinned to ROS1 Kinetic/Melodic on Ubuntu 16.04/18.04 with no ROS2 port, which means wrestling with an EOL OS/ROS combo or doing the porting work yourself. No commits since October 2023 and the usual SLAM-repo fragility around exact PCL/Ceres version compatibility, so expect to burn time on the build before you get to the algorithm.

We read the README, the directory tree and a key source file in October 2026. Nobody pays to be listed. How we judge →

search the same ground:

// at a glance

Stars
2,449
Forks
829
Language
C++
License
NOASSERTION
Last commit
19 Oct 2023

// nearby

If A-LOAM isn't quite it.

All repos →