Livox Slam (with lio+ closed loop detection optimization)

The theme ：Livox radar LIO+ Closed loop detection optimization

Open source address ：LiDAR SLAM

The open source is Livox Radar realizes an integrated and ready to use LiDAR Inertial odometer system .
The front end is based on the open source odometer framework LIO-Livox Get odometer information , In addition, more constraints and functions are added to enhance the performance of mapping .
In another ros Function pack Livox-Localization It provides a simple positioning module based on laser radar map .
summary ：
LIO-Livox Used in odometer section , ScanContext For closed-loop part and interactive_slam For map editing

1 characteristic

1.1 Closed loop detection

stay LCD part , Use ScanContext Perform closed loop detection , Then add the closed-loop constraint to the optimization solver . With the help of LCD modular , It can narrow the gap and compensate the cumulative translation error of scanning matching in large environments .

1.2 Ground restraint

In order to avoid along Z Axis drift ,LOAM（ Lidar odometer and mapping ） General problems in the algorithm , We bound the ground （ Plane detection is made by LIO-Livox complete ） As hdl_graph_slam Add to the optimization solver .

1.3 From multiple Rosbag Map merge

In practice , We usually collect environmental data many times , So we need a function to integrate from multiple rosbag Map information of . In this implementation , We use GPS Information （ Optional ） Collected at different times LiDAR The scan is converted to a coordinate system , And then use ScanContext and ICP Find the fine matching result . meanwhile , We maintain a lifelong optimization vector at the back end to store the previous key frames . In this way , We can connect scans in different sets .

1.4 GPS factors

We also offer GPS Factors to help map . This is only an optional choice . without GPS Information , The whole system can also work .

1.5 Compatible mapping editor interactive_slam

interactive_slam Is an open source based on 3D LiDAR Mapping framework for . We added a feature , In order to interactive_slam Output mapping results in a compatible format , Through this function , You can easily edit the mapping results and get more accurate LiDAR Map . meanwhile , If you are in Livox There is a camera next to it , We also provide a tool mmdection And simple LiDAR Point projection removes the function of virtual moving objects .

2 Compile operation

2.1 The environment required

Ubuntu (tested on 16.04 and 18.04)

ROS (tested with Kinetic and Melodic)

Eigen

Ceres Solver

PCL

livox_ros_driver

libsuitesparse

2.2 compile

cd ~/catkin_ws/src

git clone https://github.com/PJLab-ADG/LIO_Livox_SLAM.git

cd ..

catkin_make

2.3 function

nothing GPS data

cd ~/catkin_ws

source devel/setup.bash

roslaunch livox_mapping livox_mapping.launch save_path:="PATH_TO_SAVE_SLAM_POSE_RESULT"

Yes

cd ~/catkin_ws

source devel/setup.bash

roslaunch livox_mapping livox_mapping.launch useRTK:="true" save_path:="PATH_TO_SAVE_SLAM_POSE_RESULT"

Another terminal ：

cd ~/catkin_ws

source devel/setup.bash

roslaunch livox_odometry livox_odometry.launch save_path:="PATH_TO_SAVE_ODOM_RESULT_in_txt"

Another terminal , Play bag：

rosbag play YOUR_ROSBAG.bag

2.4 Multi package merge

There are many. bag package （ Same scene ）, Can be in the first odometry Restart at the end of the thread odometry Thread and play the next package .
In the terminal of odometer thread , Use Ctrl+C Close the old odometer thread .
In the terminal of odometer thread , Use the following command to start a new odometer thread

roslaunch livox_odometry livox_odometry.launch save_path:="PATH_TO_SAVE_ODOM_RESULT_in_txt"

2.5 Author's data set

The author provides several tests collected in an industrial park in Shanghai Data sets ,
Baidu online disk download ：
Link：https://pan.baidu.com/s/17ElBOWiFVr68975FtXY8ZA
Password：pjop

2.6 Docker Mirror image

For fast and convenient testing , More recommended docker
Pull ：

docker pull siyuanhuang95/livox_slam:release

perform docker Run the command . Use... As needed -v take docker Bind to local computer .

docker run --gpus all -it -p 3316:22 --device=/dev/dri --group-add video --volume=/tmp/.X11-unix:/tmp/.X11-unix  --env="DISPLAY=$DISPLAY" siyuanhuang95/livox_slam:release /bin/bash

allow Docker Medium RVIZ Host computer PC Run in . In the local machine ：

xhost +

Reference resources ：https://github.com/PJLab-ADG/Livox-Mapping/blob/main/README.md