Normalizing azimuth angle with wrong offset [dual_return_outlier_filter]

[yucedagonurcan]

Hello,

autoware.universe/sensing/pointcloud_preprocessor/src/outlier_filter/dual_return_outlier_filter_nodelet.cpp

Line 180 in 6745976

azimuth_diff = azimuth_diff < 0.f ? azimuth_diff + 36000.f : azimuth_diff;

Doesn't it need to be 360000.f? It looks like there is this error every time we try to normalize the angle.

[badai-nguyen]

Hello @yucedagonurcan
I think the line is to make sure the difference of adjacent points in a ring azimuth_diff being in range [0.0f 36000.0f].
Could you describe more details about the error have you get?

[yucedagonurcan]

@badai-nguyen , I am guessing that we need to fix the difference between [0, 360) not [0, 36).

[badai-nguyen]

@yucedagonurcan Hi, I am sorry for misunderstanding your question.

Doesn't it need to be 360000.f? It looks like there is this error every time we try to normalize the angle.

Because the current LiDAR's azimuth represents angle in hundredths of degree so that it should be 36000.0f.

[yucedagonurcan]

Okay, it's a fair approach if we know it is hundredths of degree

But if we are parsing the sensor data from the Ouster, user tend to calculate the azimuth value as millidegrees which is the format of the incoming data.

@badai-nguyen maybe the best approach to make it a parameter rather than fix value? What do you think?

[badai-nguyen]

@yucedagonurcan yes, I totally agree that it is better to parameterize that. I will fix it.

[badai-nguyen]

I close this issue because I created an related one here #1127