dynamic_point_indexes

[narutojxl]

Hi author, Thanks for your hard work ! I have a question about dynamic_point_indexes returned by calcDescrepancyAndParseDynamicPointIdxForEachScan() function. We use calcDescrepancyAndParseDynamicPointIdx() to parse dynamic points for each kscan.
https://github.com/irapkaist/removert/blob/828753fed2aeb714b1af513eb06fd353cd4e0b5d/src/Removerter.cpp#L568 Every point in map_local_curr_, whose index in map_local_curr_ is stored in the corrresponding row and col pixel of map_rimg_ptidx. map_local_curr_ is the map points converted into the kscan's frame using transformGlobalMapToLocal(idx_scan);

Question: It seems that when we call auto [map_rimg, map_rimg_ptidx] = map2RangeImg(map_local_curr_, kFOV, _rimg_shape); to calculate every point's index of map_local_curr_ point cloud, should be index of map_global_curr_ point cloud, not in index of map_local_curr_ point cloud ?

• first reason:
  Because when we for_each scan Done, we use std::set<> to remove repeated indexes makes sense, otherwise, even if dynamic_point_indexes has duplicated number, they maybe different. For example point_a index comes from map_global_curr_ converted into mscan, point_b index comes from map_global_curr_ converted into nscan, their index are the same , but they are different points in map. Similarly, different index points may be the same point in the map. https://github.com/irapkaist/removert/blob/828753fed2aeb714b1af513eb06fd353cd4e0b5d/src/Removerter.cpp#L582-L584
• second reason: When we parseDynamicMapPointcloudUsingPtIdx(), we use map_global_curr_, the line 441.

I' m not sure i totally understand code correctly and not miss something, look forward to disscuss with you ! Thanks for your time! https://github.com/irapkaist/removert/blob/828753fed2aeb714b1af513eb06fd353cd4e0b5d/src/Removerter.cpp#L436-L448

[narutojxl]

Sorry, it's a mistake, pcl::transformPointCloud(*map_global_curr_, *map_local_curr_, base_pose_inverse); , input point cloud and output point cloud's index do not change, so the code is correct, close it now.