SLAM Algorithms¶
The benchmark evaluates nine state-of-the-art 3D LiDAR-based SLAM algorithms:
- A-LOAM: An advanced implementation of LOAM (LiDAR Odometry and Mapping)
- LeGO-LOAM-BOR: A fork of LeGO-LOAM with good software engineering practices to make the code more readable and efficient
- LIORF: A fork of LIO-SAM which removes the feature extraction module and makes it easier to adapt other sensors
- DLIO: A lightweight LiDAR-Inertial Odometry (LIO) algorithm with a coarse-to-fine approach in constructing continuous-time trajectories for precise motion correction
- VineSLAM: A localization and mapping algorithm designed for challenging agricultural environments
- KISS-ICP: An LiDAR Odometry ICP pipeline with the KISS principle (Keep It Simple and Scalable)
- GLIM: An versatile and extensible range-based 3D mapping framework
- Kinematic-ICP: An LiDAR Odometry ICP pipeline with kinematic constraints for wheeled robots
- MOLA-LO: A modular optimization framework for localization and mapping using LiDAR Odometry (LO)
| SLAM Algorithm | Code Repository | ROS Version | VLP-16 | OS1-64 | RS-5515 | Mid-360 | IMU | Wheel Odom | Loop Closure | Year |
|---|---|---|---|---|---|---|---|---|---|---|
| LOAM | A-LOAM | Noetic (ROS 1) | X | X | X | - | - | - | - | 2014 |
| LeGO-LOAM | LeGO-LOAM-BOR | Noetic (ROS 1) | X | X | X | - | - | - | X | 2018 |
| LIO-SAM | LIORF | Noetic (ROS 1) | X | X | X | - | X | - | X | 2020 |
| DLIO | Official | Noetic (ROS 1) | X | X | X | X | X | - | - | 2022 |
| VineSLAM | Official | Humble (ROS 2) | X | X | X | X | X | X | - | 2022 |
| KISS-ICP | Official | Humble (ROS 2) | X | X | X | X | - | - | - | 2023 |
| GLIM | Official | Humble (ROS 2) | X | X | X | X | X | - | X | 2024 |
| Kinematic-ICP | Official | Humble (ROS 2) | X | X | X | X | - | X | - | 2024 |
| MOLA-LO | Official | Humble (ROS 2) | X | X | X | X | - | - | - | 2024 |