lidar_SLAM
LIDAR SLAM for Autonomous Vehicles
Building command
catkin_make --only-pkg-with-deps lidar_slam
TO-DO List
- Kitti dataset preparation
- Project framework design with ROS
- Front end lidar odometry with Lidar NDT and voxel filter
- Front end refactoring
- Sensors time synchronization algorithm
- Add point cloud visualizer based on PCL
- Lidar odometry accuracy evaluation with evo
- Point cloud distortion correction/motion compensation algorithm
- Module design and refactoring
- Data pre-process
- Front end lidar odometry
- Back end graph optimization
- Loop closure
- Visualization
- Back end optimization with g2o and Ceres
- Loop closure
- Front end extension with A-LOAM
- Localization with global map
- IMU state model derivation
Test Environment
- Ubuntu 18.04
- ROS Melodic
- PCL 1.8
G2O Installation
use the zip file in the third_party
directory
cd build
cmake ..
make -j8
sudo make install
One error I met while building this project with G2O:
cholmod library not found. It is solved by installing the dependency
libsuitesparse-dev
and re-install the g2o library
sudo apt-get install libsuitesparse-dev
Paper TO-DO List
- Robust and Precise Vehicle Localization Based on Multi-Sensor Fusion in Diverse City Scenes
- LiDAR Inertial Odometry Aided Robust LiDAR Localization System in Changing City Scenes