liosam_liauto

Achieving relocalization and global real-time localization based on LIO-SAM in point cloud maps
sample video preview:无人驾驶课题
 视频教程视频中“projection::UtmProjector projector(Origin({37.528444, 122.0780557}));“是当时建立地图时GPS获取的初始坐标。

Install

System Require:
Ubuntu(tested with 18.04)
ROS(tesed with melodic)

1. git clone this repository to your ros workspace,Run in terminal:

cd path\to\your\ros_worksapce
git clone https://github.com/1475015695/liosam_liauto.git

2. Install dependency

ROS packages:

sudo apt-get install -y ros-melodic-navigation
sudo apt-get install -y ros-melodic-robot-localization
sudo apt-get install -y ros-melodic-robot-state-publisher

GTSAM packages：

sudo add-apt-repository ppa:borglab/gtsam-release-4.0
sudo apt install libgtsam-dev libgtsam-unstable-dev

3. compile this repository

cd ..
catkin_make

Congratulations！ if no error messages comeout :), After compile, you may need execute

source devel/setup.bash

to let system found our executable porgram.

Usage

1. Prepare map

Same as LIO-SAM, to build map we need Lidar、IMU、GPS( optinal ), dont be warry my bro, I provided an example ros bag to test our program. Moreover, I also provided built map file( incase you meet some problem or test in stages :)
Edit save map dir in config/params.yaml at line 45

 savePCDDirectory: "mapDir"

Then run

#in terminal
roslaunch lio_sam run_ori.launch
#in another terminal
rosbag play path/to/your/rosbag.bag

After bag play end. ctrl + c to stop roslaunch process, check /home/mapDir, you will see there are many pcd files, and folder be like:

-mapDir
--CornerMap.pcd                        #whole Corner points
--GlobalMap.pcd                        #whole Surface and Corner points
--SurfMap.pcd                          #whole surface points
--trajectory.pcd                       #move trajectoies
--transformations.pcd                  #tranform information(current/ori axis to global axis)
--keyMap                               #keyMaps, help to achieve relocation
---subCorner0.pcd                      #keyCornelFrame
---......
---subSurface0.pcd                     #keySurfaceFrame
---......

Congratulations！bro, its hard for most people to get above steps done :)
I highly recommended make a copy of final map( every time you runlaunch, map will be overwrite, may lost our final map)

2. Drow and load road map

This video is all we need, we can imitate to genrate our road map, here I name it "new_lanelet2_maps.osm" and put it in "/roscode/map_copy/", edit laneletFilePath in config/params.yaml at line 105. \

laneletFilePath: "/roscode/map_copy/new_lanelet2_maps.osm"

OK! let check wether everything is fine:

#in terminal
roslaunch lio_sam liauto.launch

If you see white points cloud map and red lane map, great! you have done above steps correctly!
Come bro, we almost done :o\

3. Relocatlization

After connect your sensor to ros, run:

#in terminal
roslaunch lio_sam liauto.launch

liauto will do relocatlization use current GPS message, but for convince popurse, if you known where you are, you can click 2d pose estimated in Rcviz and mark your probably posetion in map.

4. Realtime locatlization

After relocalization, liauto will keep track our postion in map, and output coordinate in terminal, you can use coordinate in "/currPose" message. Have Fun!

5. Material

LiDAR_IMU_Init # calib Lidar and IMU. Thanks Contributors: Fangcheng Zhu 朱方程， Yunfan Ren 任云帆， Wei Xu 徐威， Yixi Cai 蔡逸熙 :)
LIO-SAM # base project. Thanks TixiaoShan :)
Lanelet2 # load and use Lanelet map. Thanks Poggenhans, Fabian and Pauls, Jan-Hendrik and Janosovits, Johannes and Orf, Stefan and Naumann, Maximilian and Kuhnt, Florian and Mayr, Matthias :)
LIO-SAM-DetailedNote # DetailedNote for lio-sam. Thanks smilefacehh Tao Lu :)
My paper #我的毕业论文，里面有详细的原理介绍 :) 也谢谢我自己 :)

gary-robotics/liosam_liauto