[pcl::VoxelGrid::applyFilter] Leaf size is too small for the input dataset. Integer indices would overflow.

[hmellor]

I am getting the following warning/error repeatedly when using the calibration tool:

[pcl::VoxelGrid::applyFilter] Leaf size is too small for the input dataset. Integer indices would overflow.

I am using the following configuration:

• ROS noetic
• Ubuntu 20.04LTS

• Velodyne VLP-16

  (In order to compile on Ubuntu 20.04LTS I had to change the compiler from c++11 to c++14 in CMakeLists.txt, but then it appeared to compile correctly)

[chennuo0125-HIT]

because coverage of your cloud is too large, you can increase leaf size of the filter https://github.com/chennuo0125-HIT/lidar_imu_calib/blob/96f7a9d897e41eb95ad71c4ecbc87a3d0f2cccb7/src/lib/calibExRLidar2Imu.cpp#L14 as follows:

downer_.setLeafSize(0.2, 0.2, 0.2);

[hmellor]

Thanks for the quick response, that appears to have made the warnings go away.

I have another question, does the calibration run until all LiDAR msgs are added or until the solution converges?

[chennuo0125-HIT]

this method give Least square solution, so if you give more lidar msgs, you can get better result. for my experience, lidar msgs are enough collected during one minutes.

[hmellor]

Ok, thank you for your help

[ellieko]

@chennuo0125-HIT Hi, thanks for sharing the solution. I have one question though, how do we know appropriate leaf size ? I'm still getting the same error with the leaf size of 0.3, as it's shown below.

total lidar buffer size 810, imu buffer size 40479
constraints size 808
[pcl::VoxelGrid::applyFilter] Leaf size is too small for the input dataset. Integer indices would overflow.