Error in pose estimation when use vlp_16

[BamsaAhmed]

Hi , you recommended to use the FLOAM for pose estimation, I used my own dataset and use FLOAM to estimate the pose and set the scan line = 16 . But the pose estimation is error because it grew up (in z-axes) as shown below. Could you prese nt to me any advice?

[EPVelasco]

Hi, the image has not been uploaded to github for review. Try to upload it again.

[BamsaAhmed]

Thank you for quick response, the image is reloaded.[ ]

[EPVelasco]

Hello, sometimes FLOAM tends to estimate odometry less accurately when not enough points are available at the edges or surface of the current point cloud. You can adjust the minimum distance at which the FLOAM algorithm uses the point cloud data. This option is found in line 23. You can also change the resolution of the map voxelization, located on line 24. Reducing this value will provide a higher resolution. Note, however, that increasing the resolution may increase the time required for pose optimization. I would also like to know if launching the FLOAM algorithm shows you any message in the terminal.

[BamsaAhmed]

Hi I changed the maximum and minimum distance value as your advice, the estimation has been improved but still it is not accurate, i think the problem is in the vertical view angle which is not considered in the code. So could you please have information on how to set it?

‫في الأربعاء، 15 نوفمبر 2023 في 7:21 م تمت كتابة ما يلي بواسطة ‪Edison Velasco Sánchez‬‏ @.***‬‏>:‬

Hello, sometimes FLOAM tends to estimate odometry less accurately when not enough points are available at the edges or surface of the current point cloud. You can adjust the minimum distance at which the FLOAM algorithm uses the point cloud data. This option is found in line 23 https://github.com/EPVelasco/floam/blob/master/launch/floam_experimets.launch#L23. You can also change the resolution of the map voxelization, located on line 24 https://github.com/EPVelasco/floam/blob/master/launch/floam_experimets.launch#L24. Reducing this value will provide a higher resolution. Note, however, that increasing the resolution may increase the time required for pose optimization. I would also like to know if launching the FLOAM algorithm shows you any message in the terminal.

— Reply to this email directly, view it on GitHub https://github.com/EPVelasco/lidar-camera-fusion/issues/14#issuecomment-1812850159, or unsubscribe https://github.com/notifications/unsubscribe-auth/AZHQ4Z24TSQJKUPORZIV4S3YETTYBAVCNFSM6AAAAAA7MDKTWKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJSHA2TAMJVHE . You are receiving this because you authored the thread.Message ID: @.***>

[EPVelasco]

Hello, you could use the original F-LOAM algorithm. This is to know if the error is in the fork that I have created. Another thing to check is if the sensor is generating the point cloud correctly. It is possible that when you have an estimation error you do not have a point cloud generated by some visual occlusion of the sensor.