Lidar-IMU-Localization

This repository is a Lidar-IMU Localization System with Prior Map Constraint and Lio Constraint for 3D LiDAR. The system is developed based on the open-source odometry framework LIO-Livox. And the feature extract moudle is implemented based on LIO-SAM .

Mapping Moudle
- A Modified FeatureExtract Function adapt for traditional spinning lidar,such as velodyne,ouster,robosense etc. ;
- A Modified Tightly coupled Lidar-imu laserodometry LIO-Livox-modified;
Localization Moudle
- A Lidar-IMU Localization System with Prior Map Constraint and Lio Constraint for 3D LiDAR;
- Three IMU_Mode: 0-without using IMU, 1-loose couple IMU and Lidar, 2-tightly coupled IMU and LiDAR;
- Automatic switch Map-Location mode and LIO-Location mode;

demo

A test video of the dataset can be found on BiliBili

UrbanNavDataset test video can be found here

Prerequisites

Ubuntu (tested on 18.04)
ROS (tested with Melodic)
Eigen
Ceres Solver
PCL
livox_ros_driver

Suitesparse

sudo apt-get install libsuitesparse-dev

Compilation

cd ~/catkin_ws/src
git clone https://github.com/chengwei0427/Lidar_IMU_Localization
cd ..
catkin_make

Run with bag

(1) generate global map with LIO-SAM-modified

roslaunch GC_LOAM run.launch

rosbag play yourbagname.bag --clock

rosserve call /save_map

(2) run localization with global map and your test bag

rosbag LIO_Localization run_loc.launch

rosbag play yourbagname.bag --clock

Set initial pose in rviz

Notes

The current version of the system is just a demo and we haven't done enough tests.

There are some parameters in params.yaml files:

IMU_Mode: choose IMU information fusion strategy, there are 3 modes:
- 0 - without using IMU information, pure LiDAR odometry, motion distortion is removed using a constant velocity model (added 2022-09-16)
- 1 - using IMU preintegration to remove motion distortion (added 2022-09-19)
- 2 - tightly coupling IMU and LiDAR information (added 2022-09-27)

TODO

Acknowledgements

Thanks for LOAM,LIO_SAM ,LIO-Livox.

li9616/Lidar_IMU_Localization