You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
After carefully comparing the codes and the paper, I found:
init_plane function, calculate the plane's (n,q) and cov, when initalizating or updating the octree, which is the same as paper's description.
update_plane function, calculate the plane's (n,q) without cov. This function is only called in updateOctoTree when (the node is the largest layer & not a plane & no-updating the cov). But as the paper descirbed, there should be a function that "check the plane consistancy, and then only update the (n,q) if the same plane. The update_plane doesn't do this.
Thus, I think there are two issues that incosistant with the paper:
The update_plane function should check plane consistancy first, but not directly re-calcuate the paramter. I can still find some "old" variables here:
The update_plane should be called in another place (Line 402 in UpdateOctoTree:
if (new_points_num_ > update_size_threshold_) {
if (update_cov_enable_) {
init_plane(temp_points_, plane_ptr_);
}
//** there should be "update_plane" if `update_cov_enable_` is false. **//
new_points_num_ = 0;
}
In this cases, the cov do not need to be update, but only update the (n,q).
Could someone explain what's the motivation that the code is not consistant with the original paper. Thanks.
The text was updated successfully, but these errors were encountered:
Hi, thanks for releasing the code.
After carefully comparing the codes and the paper, I found:
init_plane
function, calculate the plane's (n,q) and cov, when initalizating or updating the octree, which is the same as paper's description.update_plane
function, calculate the plane's (n,q) without cov. This function is only called inupdateOctoTree
when (the node is the largest layer & not a plane & no-updating the cov). But as the paper descirbed, there should be a function that "check the plane consistancy, and then only update the (n,q) if the same plane. Theupdate_plane
doesn't do this.Thus, I think there are two issues that incosistant with the paper:
update_plane
function should check plane consistancy first, but not directly re-calcuate the paramter. I can still find some "old" variables here:in Line 247 of "voxel_map_util.hpp"
update_plane
should be called in another place (Line 402 inUpdateOctoTree
:In this cases, the cov do not need to be update, but only update the (n,q).
Could someone explain what's the motivation that the code is not consistant with the original paper. Thanks.
The text was updated successfully, but these errors were encountered: