Laser Odometry and Localization
Add localization support for LOAM/LeGO-LOAM, referenced from 8.2.3 Localization and Map Merging in [1].
Compared to the previous section, the difference is that the localization does not process individual scans but stacks a number of scans for batch processing. Thanks to the high-accuracy odometry estimation, scans are registered precisely in a local coordinate frame where drift is negligible over a short period of time (a few seconds).
Localization test(running at TX2) under point cloud map built by A-LeGO-LOAM[2].
Map bulit by lol using corner point cloud.
- /surf: surf point cloud extracted by Laser Odometry
- /corner: corner point cloud extracted by Laser Odometry
- /outlier: outlier point cloud extracted by Image Projection
- /odom/lidar: odometry estimeted by Laser Odometry
- global map pcd file: built by A-LeGO-LOAM
- /tf: map -> odom
- /currenct_pose: PoseStamped
- ...
- Run the launch file:
roslaunch lol lol.launch
- Play existing bag files test_0515.bag:
rosbag play test_0515.bag --clock --topics /lslidar_point_cloud
- Zhang J. Online Lidar and Vision based Ego-motion Estimation and Mapping[D]. , 2015.
- https://github.com/jyakaranda/A-LeGO-LOAM