A question about the detail when pre-process KITTI Velodyne data

[LongruiDong]

Hi~
Thanks for your great work.

I am quite new with KITTI Velodyne and confused about this "error" :

an error of 0.22° in the Velodyne sensor

Does this angle error occur in all KITTI Odometry training sequences (00-10)?
I also read the relevant paper[46], but still don't know how to correct it...

Can you explain more in detail?

Thanks in advance~

[SRainGit]

Hi~
Thanks for your great work.

I am quite new with KITTI Velodyne and confused about this "error" :

an error of 0.22° in the Velodyne sensor

Does this angle error occur in all KITTI Odometry training sequences (00-10)?
I also read the relevant paper[46], but still don't know how to correct it...

Can you explain more in detail?

Thanks in advance~

Hi, Thanks.
0.22° is one of the intrinsic parameters. The scanning model is shown in the figure. The actual scanning lines in KITTI are higher than they are supposed to be by 0.22°. So, rotating all the points up by this angle is what we did.

[wwtinwhu]

Hi~
Thanks for your great work.
I am quite new with KITTI Velodyne and confused about this "error" :

an error of 0.22° in the Velodyne sensor

Does this angle error occur in all KITTI Odometry training sequences (00-10)?
I also read the relevant paper[46], but still don't know how to correct it...

Can you explain more in detail?
Thanks in advance~

Hi, Thanks.
0.22° is one of the intrinsic parameters. The scanning model is shown in the figure. The actual scanning lines in KITTI are higher than they are supposed to be by 0.22°. So, rotating all the points up by this angle is what we did.

Thanks for explaining this. That is to say every point will be transformed with a 0.22 degree rotation in roll？if so, why call this intrinsic parameter? This can be thought of the overall transformation between lidar frame and the gps/imu's frame. And, Have you done the experiment that not to preprocess the point cloud, how about the average translation error in this situation?

[SRainGit]

Hi~
Thanks for your great work.
I am quite new with KITTI Velodyne and confused about this "error" :

an error of 0.22° in the Velodyne sensor

Does this angle error occur in all KITTI Odometry training sequences (00-10)?
I also read the relevant paper[46], but still don't know how to correct it...

Can you explain more in detail?
Thanks in advance~

Hi, Thanks.
0.22° is one of the intrinsic parameters. The scanning model is shown in the figure. The actual scanning lines in KITTI are higher than they are supposed to be by 0.22°. So, rotating all the points up by this angle is what we did.

Thanks for explaining this. That is to say every point will be transformed with a 0.22 degree rotation in roll？if so, why call this intrinsic parameter? This can be thought of the overall transformation between lidar frame and the gps/imu's frame. And, Have you done the experiment that not to preprocess the point cloud, how about the average translation error in this situation?

Hi, it's still an intrinsic parameter. Because it is an internal distortion, not a rotation.
My experiments show that 0.22 degree of correction did lower the error in sequence 01 (highway scene) but had a minor effect on other sequences. One friend of mine said his method achieved higher error after the correction. Maybe it depends. But the distortion is a fact. I will try to add this part to my articles. Thanks for your questions.

[wwtinwhu]

Hi~
Thanks for your great work.
I am quite new with KITTI Velodyne and confused about this "error" :

an error of 0.22° in the Velodyne sensor

Does this angle error occur in all KITTI Odometry training sequences (00-10)?
I also read the relevant paper[46], but still don't know how to correct it...

Can you explain more in detail?
Thanks in advance~

Hi, Thanks.
0.22° is one of the intrinsic parameters. The scanning model is shown in the figure. The actual scanning lines in KITTI are higher than they are supposed to be by 0.22°. So, rotating all the points up by this angle is what we did.

Thanks for explaining this. That is to say every point will be transformed with a 0.22 degree rotation in roll？if so, why call this intrinsic parameter? This can be thought of the overall transformation between lidar frame and the gps/imu's frame. And, Have you done the experiment that not to preprocess the point cloud, how about the average translation error in this situation?

Hi, it's still an intrinsic parameter. Because it is an internal distortion, not a rotation.
My experiments show that 0.22 degree of correction did lower the error in sequence 01 (highway scene) but had a minor effect on other sequences. One friend of mine said his method achieved higher error after the correction. Maybe it depends. But the distortion is a fact. I will try to add this part to my articles. Thanks for your questions.

Oh, I get that! It is intrinsic parameter indeed. By the way, do you know how to calculate this calibration angle? I want to do some experiments on this preprocessing.

[SRainGit]

It seems this is a hot topic. But we are trying to make the verification of the 0.22° more scientific and convincing. It is not proper to release the idea here now. Sorry about that. @wwtinwhu

[wwtinwhu]

It seems this is a hot topic. But we are trying to make the verification of the 0.22° more scientific and convincing. It is not proper to release the idea here now. Sorry about that. @wwtinwhu

Ok, Thanks anyway! Looking forward to your new paper! @SRainGit

[SRainGit]

I will make a revision on my preprint mainly for the intrinsic parameters, but no new papers I think.

[YuePanEdward]

Great work.
I wonder if the updated version of the preprint concerning the intrinsic parameters would be published?

[SRainGit]

Great work.
I wonder if the updated version of the preprint concerning the intrinsic parameters would be published?

Thanks. Yes, I will update the intrinsic part into the new version. Perhaps at the end of this month. But it's just a simple idea, not a full intrinsic calibration solution.