LIO-SAM

An open-source tightly coupled lidar-inertial odometry and mapping system built on factor-graph optimization.

LIO-SAM is a lidar-inertial SLAM system by Tixiao Shan, Brendan Englot, Daniela Rus, and colleagues, published at IROS 2020; Shan and Englot had previously built LeGO-LOAM. It formulates localization and mapping as a factor graph (a graph model that treats each measurement as a constraint and jointly optimizes a sequence of poses), supporting four types of factors: IMU preintegration, lidar odometry, GPS, and loop closure. IMU preintegration both deskews the point cloud and provides an initial guess for registration. To stay real-time, it only selects keyframes and registers each new keyframe against a local map made from a fixed number of nearby historical keyframes, rather than matching against the global map. Its back end is built on the GTSAM library. The code is open source under a BSD-3 license, with a ROS 1 version and a ROS 2 branch.

ExampleThe README notes that LIO-SAM only works with a 9-axis IMU that can output roll, pitch, and yaw, recommended at 200 Hz or higher; when using an Ouster lidar, its built-in 6-axis IMU is not sufficient, so a separate external 9-axis IMU is needed.

Also called

LiDAR Inertial Odometry via Smoothing and Mapping

Related

LiDAR-Inertial Odometry · Factor Graph Optimization · IMU Preintegration · GTSAM (Georgia Tech Smoothing and Mapping) · Loop Closure Detection · LOAM (LiDAR Odometry and Mapping)

Sources

LIO-SAM: Tightly-coupled Lidar Inertial Odometry via Smoothing and Mapping (IROS 2020, arXiv 2007.00258)
GitHub: TixiaoShan/LIO-SAM