It's not working with Livox mid-360, even after I am using PointCloud2 format.

[naveenkumar2208]

[LarryDong]

Hi, can you provide more detials? I'm using mid360 and meet some problems. I'm not sure if we have the same issue. Thanks.

[JulesThuillier]

Hello @LarryDong ! I would like to use the Livox MID 360 and Livox Hap, were you happy with the results obtained with Lego-LOAM ? Did you use the internal IMU of the MID360 ? And of course if you have to time to share a few explanation on the modifications that would be awesome ! =)

[LarryDong]

Hello @LarryDong ! I would like to use the Livox MID 360 and Livox Hap, were you happy with the results obtained with Lego-LOAM ? Did you use the internal IMU of the MID360 ? And of course if you have to time to share a few explanation on the modifications that would be awesome ! =)

Hi, my suggestion is not to use mid360. Here are some findings:

• Mid360's point cloud is really sparse comparing to other lidar, so you need to adjust the parameters a lot which is different due to the scene and your applications;
• You can re-calculate the "ring" information based on vertical angles and save them into different rings. (Here is my code that convert mid360's CustomMsg into ros PointCloud2 with XYZ+Ring https://github.com/hust-ddc-slam/mid360-utils/tree/main/msg_converter) . However, mid360's vertical FoV is small (below the horizontal plan) so there are only 2-5 "equivalent" scans on the ground, making the ground segmentation really difficult; Tilted installation seems no help because the ground segmentation based on lego-loam is not suitable. Maybe you could install the lidar up side down (but I never tried) to make more ground points if the surronding scene is not so important.
• Uisng internal IMU is okay. The only change may be the acceleration scale (for mid360 is in 'g' not 'm/s2').

So in all, do not try to use mid360 if possible :< For Livox Hap, no test.

Good luck.