beltransen / velo2cam_calibration

Automatic Extrinsic Calibration Method for LiDAR and Camera Sensor Setups. ROS Package.
http://wiki.ros.org/velo2cam_calibration
GNU General Public License v2.0
684 stars 191 forks source link

Will the result being influenced or not if I put my lidar vertically and scan vetically #54

Open yulan0215 opened 1 year ago

yulan0215 commented 1 year ago

Hi:

Recently I worked in the camera lidar calibration with this calibration tool. However, the problem was that if I change the position of my LiDAR sensor, then my LiDAR sensor will be vertically mounted.

Then the Ring of point cloud will be vertical distributed. The question is: if my LiDAR sensor scanns vertically whether it will influence the result of finding the 3D corner in LiDAR?

Normally in the calibration, like the example in this calibration tool, the LiDAR was horizontal and scanns horizontally...

Or if neccessay I can also rotate the board...

cguindel commented 1 year ago

This is an excellent question. The LiDAR segmentation branch should be able to detect the circles regardless of the LiDAR orientation; however, it can make it difficult to determine their order, which is required to associate them with the corresponding ArUco marker.

I think it would be indeed a good idea to rotate the calibration pattern along with the sensor so that the circles maintain their relative positions in the point cloud (i.e., the top-left circle is in the top-left position in LiDAR coordinates, and so on). This means that you do not need to apply exactly the same rotation to the pattern as to the sensor, but if, for example, the latter is to be rotated 90º, the pattern will need a rotation greater than 45º in that direction.

I hope my answer helps you. If possible, please come back to us with your feedback.