This package allows to obtain the extrinsic calibration between two PointClouds with the help of the NDT algorithm.

The multi_lidar_calibrator node receives two PointCloud2 messages (parent and child), and an initialization pose. If possible, the transformation required to transform the child to the parent point cloud is calculated, and output to the terminal.

1. You'll need to provide an initial guess, otherwise the transformation won't converge.

2. In a sourced terminal:

Using rosrun

rosrun multi_lidar_calibrator multi_lidar_calibrator _points_child_src:=/lidar_child/points_raw _points_parent_src:=/lidar_parent/points_raw _x:=0.0 _y:=0.0 _z:=0.0 _roll:=0.0 _pitch:=0.0 _yaw:=0.0

Using roslaunch

roslaunch multi_lidar_calibrator multi_lidar_calibrator.launch points_child_src:=/lidar_child/points_raw points_parent_src:=/lidar_parent/points_raw x:=0.0 y:=0.0 z:=0.0 roll:=0.0 pitch:=0.0 yaw:=0.0

3. Play a rosbag with both lidar data /lidar_child/points_raw and /lidar_parent/points_raw

4. The resulting transformation will be shown in the terminal as shown in the Output section.

5. Open RViz and set the fixed frame to the Parent

6. Add both point cloud /lidar_parent/points_raw and /points_calibrated

7. If the algorithm converged, both PointClouds will be shown in rviz.

1. Child Point cloud transformed to the Parent frame and published in /points_calibrated.
2. Output in the terminal showing the X,Y,Z,Yaw,Pitch,Roll transformation between child and parent. These values can be used later with the static_transform_publisher.

transformation from ChildFrame to ParentFrame
This transformation can be replicated using:

rosrun tf static_transform_publisher 1.7096 -0.101048 -0.56108 1.5708 0.00830573  0.843 /ParentFrame /ChildFrame 10

The figure below shows two lidar sensors calibrated by this node. One is shown in gray while the other is show in blue. Image obtained from rviz.