Skip to content

[IROS 2021] CLINS: Continuous-Time Trajectory Estimation for LiDAR-Inertial System

License

Notifications You must be signed in to change notification settings

APRIL-ZJU/clins

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

clins

CLINS : Continuous-Time Trajectory Estimation for LiDAR-Inertial System

CLINS is a highly-accurate continuous-time trajectory estimation framework dedicated for SLAM (Simultaneous Localization and Mapping) applications, which enables fuse high-frequency and asynchronous sensor data effectively. Now CLINS is applied in a 3D LiDAR-inertial system for evaluations.

Dependency

As a SLAMer, we are confident that you have the ability to solve the dependency by yourself.

Build

# clone clins into your catkin workspace
git clone https://github.com/APRIL-ZJU/clins.git

catkin_make

Examples

0. Recover timestamp

Some public datasets do not provide a timestamp for each point in the scan. We tried to recover the exact timestamp of each point. The method shown below is adopted in LOAM however it is not suitable when the scan is incomplete. We refer readers to the src/node/recovery_vlp16_timestamp.cpp for details.

1. Indoor Dataset from lio-mapping

The lidar msg in the original datasets does not have the time filed, we recover each point' timestamp of the scan and available here.

Before you launch the file, check the bag_path in the configuration file.

# Run the launch file:
roslaunch clins clins_offline.launch config_name:=/config/liom-bag.yaml

Before you launch the file, check the bag_path in the configuration file.

# Run the launch file:
roslaunch clins clins_offline.launch config_name:=/config/yq-data.yaml

yq-skg-around

Before you launch the file, check the bag_path in the configuration file.

# Run the launch file:
roslaunch clins clins_offline.launch config_name:=/config/lio-sam-bag.yaml

Before you launch the file, check the bag_path in the configuration file.

# Run the launch file:
roslaunch clins clins_offline.launch config_name:=/config/kaist.yaml

The estimated trajectory is saved in the directory of the provided dataset. Modify the parameters trajectory_path, bag_path, and save_cloud_path in kaist_sick_map_generation.launch, then run it!

# use the estimated trajectory to generate map with 2D LiDAR data from SICK LMS-511
roslaunch clins kaist_sick_map_generation.launch

You will get the following reconstruction result.

Run with your own data

All the above sequences are collected by VLP-16.

Note that ring field is required for cloud feature extraction. time field is required for undistortion. Make sure the lidar topic (sensor_msgs/PointCloud2 ) has both ring and time fields.

Credits

This code was developed by the APRIL Lab in Zhejiang University.

For researchers that have leveraged or compared to this work, please cite the following:

Jiajun Lv, Kewei Hu, Jinhong Xu, Yong Liu, Xingxing Zuo. CLINS : Continuous-Time Trajectory Estimation for LiDAR-Inertial System. IROS 2021. [arxiv]

The lidar feature extraction is heavily derived from LIO-SAM. The continuous-time trajectory is taken from basalt.

TODO

  • LiDAR odometry saves all lidar scan which is slow for large dataset.
  • Support livox LiDAR.
  • Marginalization LiDAR feature and IMU measurement.
  • Speed up (analytic factor).

About

[IROS 2021] CLINS: Continuous-Time Trajectory Estimation for LiDAR-Inertial System

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published