FAST-LIO2

An open-source lidar-plus-IMU odometry system from HKU that's fast and works with many lidar types.

FAST-LIO is an open-source lidar-inertial odometry system from the MARS Lab at the University of Hong Kong (Fu Zhang's group): it fuses lidar point clouds with an IMU (which measures acceleration and angular velocity) to estimate the robot's own pose in real time while simultaneously building a point-cloud map. The first version used a tightly coupled iterated extended Kalman filter. FAST-LIO2, released in July 2021, made two key changes: it no longer hand-extracts edge and plane features but registers raw points directly against the map, so it works with both spinning lidars (Velodyne, Ouster) and solid-state ones (Livox); and it maintains its map with a custom incremental k-d tree called ikd-Tree, with the paper reporting rates up to 100 Hz in large scenes. It's a common open-source baseline for localization and mapping on quadrupeds, drones, and humanoids, is licensed under GPL-2.0, and runs on ARM boards.

ExampleMounting a Livox solid-state lidar on a quadruped's back and running FAST-LIO2 gives its self-pose and a point-cloud map in real time, which is then handed to a navigation module for path planning.

Also called

FAST-LIO, Fast LiDAR-Inertial Odometry

Related

LiDAR-Inertial Odometry · LiDAR · Inertial Measurement Unit · LiDAR SLAM · LIO-SAM · Solid-State LiDAR

Sources

FAST-LIO2: Fast Direct LiDAR-inertial Odometry (arXiv:2107.06829)
hku-mars/FAST_LIO (GitHub)

As of

2021-07