liangheming/FASTLIO2_ROS2

What it solves

This project provides a ROS2 adaptation of the FASTLIO2 framework, enabling high-performance LiDAR-Inertial Odometry (LIO) with added capabilities for loop closure, relocalization, and map consistency optimization to prevent drift in robotic mapping and navigation.

How it works

The system integrates several specialized components:

• LIO Core: A refactored version of FASTLIO2 adapted for ROS2 Humble.
• Loop Closure: Uses position priors and Iterative Closest Point (ICP) for detection, with GTSAM for pose-graph optimization.
• Consistancy Optimization: Employs BLAM for small-scale scenes and HBA for large-scale scenes to refine the map.
• Relocalization: Implements a two-stage (coarse-to-fine) ICP process to determine the robot's position within a known map.

Who it’s for

Developers and robotics engineers working with ROS2 Humble on Ubuntu 22.04 who need a robust LiDAR-Inertial Odometry system with mapping and relocalization features.

Highlights

• ROS2 Humble Support: Full migration of FASTLIO2 to the ROS2 ecosystem.
• Loop Closure & PGO: Integrated pose-graph optimization to correct trajectory drift.
• Two-Stage Relocalization: Efficient coarse-to-fine ICP for fast position recovery.
• Scalable Map Optimization: Support for both small and large scene optimizations via BLAM and HBA.