Questions about the Sequence of Data Association

[zxd123]

Thanks for sharing the great work!
But, I have a question about your paper. In section IV-D 4), you mentioned that for data association, one needs to first specify a point and a matching plane, and then look for the leaf node in the octree which contains the centroid of the plane to determine the matching relationship with the data point. However, I'm a bit confused about the sequence of this process. If we don't look for the leaf node in the octree, how can we know which plane the data point matches with?
Looking forward to your answer, thank you very much.

[lewisjiang]

Hi @zxd123,

It is a little confusing indeed, because we split matching a plane from matching measurement points. The first step gives the linearization point of the plane feature, and the second step gives the "observations of the plane" from older poses, which are used in measurement model. The reasons for this are: the plane calculated from the first step is more stable than using close measurement points, and the reuse of measurement points can be avoided.

[zxd123]

Thank you very much for your reply. Looking forward to the open-source of this project.

[zxd123]

In the paper, it's mentioned that to process the data, one should first identify the plane that corresponds to a given point, and then locate the octree node containing the centroid of that plane to find the associated measurement points. However, could you please elaborate on the initial step? Specifically, once I have a set of measurement points from a frame, how is the corresponding plane for each point determined?

中文：
论文中提到了需要先通过点找到其匹配的平面，再根据平面找到八叉树中包含该平面质心的节点，再在节点中找到对应的测量点。但是我希望请教下第一步是怎么做的呢？也就是说刚拿到一帧点云后，是如何确定每个点所匹配到的平面呢？

[lewisjiang]

@zxd123

也就是说刚拿到一帧点云后，是如何确定每个点所匹配到的平面呢？

和fastlio2 一样

[zxd123]

@zxd123

也就是说刚拿到一帧点云后，是如何确定每个点所匹配到的平面呢？

和fastlio2 一样

好的好的，非常感谢