Some questions about the paper

[a-github-learner]

Thank you for your fascinating work! I have a few questions regarding the details in your paper:

How did you establish the correspondences between different robot maps in your study? For instance, considering input data such as the example below:

2. In your paper, the GICP method achieved an RRE of 0.22 and an RTE of 8.56. Could you elaborate on how you attained this level of precision?

[NothingCD]

您好，邮件已收到，尽快给您回复。

[LimHyungTae]

@a-github-learner

1. Hello, we have established Faster-PFH (faster version of FPFH) + descriptor matching.

2. We just used generic G-ICP from PCL. Have this answered your question by the way?

[a-github-learner]

Thank you for your reply! How do you establish the descriptor matching in unbalanced map? Have you used sliding windows or other methods for coarse matching?

[LimHyungTae]

I just did voxelization -> descriptor matching for all voxelized points -> pruning.