params for 128-beam LiDAR

[WSTao]

When I test with 128-beam data, the ground will be under-segmented，how to modify those parameters? thanks much！

[LimHyungTae]

Could you please give me some screenshots in rviz? I guess more than 64 channels do not cause the under-segmentation.

[WSTao]

Could you please give me some screenshots in rviz? I guess more than 64 channels do not cause the under-segmentation.

[LimHyungTae]

Thank you for your fast reply! In addition, could you show the visualized likelihood?? Did you change the parameter sensor_height to your own sensor_height?

[WSTao]

Thank you for your fast reply! In addition, could you show the visualized likelihood?? Did you change the parameter sensor_height to your own sensor_height? I modified it, but there is no obvious improvement

[WSTao]

I modified the parameter flatness_thresholds and the effect has been significantly improved, but there are still under-segmentation and over-segmentation，how is elevation_thresholds and flatness_thresholds determined?

[LimHyungTae]

If then, it denotes that most bins are rejected via elevation filters. I guess your elevation thresholds should be lower than [-1.2, -0.9984, -0.851, -0.605]. As shown in the screenshots, your sensor height is quite high....so how about change the elevation thresholds?

• you can check the status by visualizing the jsk_rviz_plugin/PolygonArray msg named as /gpf/plane

[WSTao]

If then, it denotes that most bins are rejected via elevation filters. I guess your elevation thresholds should be lower than [-1.2, -0.9984, -0.851, -0.605]. As shown in the screenshots, your sensor height is quite high....so how about change the elevation thresholds?

• you can check the status by visualizing the jsk_rviz_plugin/PolygonArray msg named as /gpf/plane

My lidar height is 1.98, I will try the method you mentioned，Thank you very much!

[LimHyungTae]

Hello there, did you solve the problem? Could you share your /gpf/plane msg please? I'm curious your situations :)

[WSTao]

Hello there, did you solve the problem? Could you share your /gpf/plane msg please? I'm curious your situations :)

likelihood.txt Thank you very much, I printed the message, but I don’t know how to analyze it

[LimHyungTae]

Oh, I mean I'd like to viz GLE as follows: You can debug what the problem is via visualized GLE.

[WSTao]

Oh, I mean I'd like to viz GLE as follows: You can debug what the problem is via visualized GLE.

[LimHyungTae]

Cool! The blue parts denote that your elevation threshold should be fixed. As you mentioned earlier, the height of the platform is 1.98, why don't you change the elevation terms, i.e. [-1.2, -0.9984, -0.851, -0.605], into [-1.46, -1.256, -1.10, -0.862] or more lower values?

[WSTao]

Thank you very much, I modified the variable, but the result is basically the same

[LimHyungTae]

Could you send me some pcds? It would be helpful for generalization of algorithm: shapelim@kaist.ac.kr . I'll check it

[WSTao]

Thank you for your patience, the pcd file has been sent to you

[MingXXI]

Cool! The blue parts denote that your elevation threshold should be fixed. As you mentioned earlier, the height of the platform is 1.98, why don't you change the elevation terms, i.e. [-1.2, -0.9984, -0.851, -0.605], into [-1.46, -1.256, -1.10, -0.862] or more lower values?

Very appreciate your helpful discussion :) Could I get some hint about how the threshold_value are set? According to the comments found in your config, "1.723 - 1.2 = 0.523 is acceptable", I'm wondering how does 1.2 calculated, and is there a proper way to set all thresholds nicely?