- scan match
- correspondence
- scan-to-map match
- map update
ICP/PLICP method is used for scan-to-scan match, and ceres is used as the nonlinear solver with point-to-line contraints.
KNN method of pcl lib is used to find point correspondence
find local optimal with random search
OpenCV function line and circle is used to update grid map.
cd catkin_ws/src
git clone [email protected]:libing64/slam2d.git
cd ..
catkin_make -DCATKIN_WHITELIST_PACKAGES="slam2d"
soure devel/setup.bash
roslaunch slam2d slam2d.launch
play your rosbag
rosbag play youbag
cartographer dataset:
https://google-cartographer-ros.readthedocs.io/en/latest/data.html
- Ubuntu20.04
- ros noetic
- OpenCV
- pcl
- Eigen
- ceres
- how to reduce the drift
- how to improve the mapping accurcy
- scan-to-map match
- keyframe selection to reduce drift
- dynamic range 2d map
- search local optimal
- record gif with byzana-record
- sparse pose adjust
- loop closure
- scan-to-scan match and scan-to-map match, Gaussian-Newton
- need high rate scan data
- 2.5D SLAM
- multi-resolution map to avoid local minima
- Rao-Blackwellized Particle filter approach
- graph-based approach
- landmark based
- Cholesky decomposition for solving sparse linar system
- loop closure
- sparse pose adjustment(just like bundle adjustment)
- ros wrapper of tinySLAM(200 lines-of-code)
- partical filter based approach
- Graph optimization -> Linear approximation of Graph optimization
- linear so no initial value needed
- grid_map https://github.com/ANYbotics/grid_map
- costmap_2d https://wiki.ros.org/costmap_2d
- OctoMap https://octomap.github.io/
- An ICP variant using a point-to-line metric
- A flexible and Scalable SLAM System with Full 3D motion Estimation
- Real-Time Loop Closure in 2D LIDAR SLAM
- Grid-based Scan-to-Map Matching for Accurate 2D Map Building