Skip to content

Measure ICP convergence with the chosen nearest neighbors - #1500

Open
Afloat16 wants to merge 1 commit into
AtsushiSakai:masterfrom
Afloat16:fix/icp-association-residual
Open

Afloat16 wants to merge 1 commit into
AtsushiSakai:masterfrom
Afloat16:fix/icp-association-residual

Conversation

@Afloat16

@Afloat16 Afloat16 commented Nov 6, 2026

Copy link
Copy Markdown

Reference issue

Related to #1453. That PR addresses transform composition.

What does this implement/fix?

The convergence residual subtracted input columns pairwise before matching them. Reordering one scan changed the residual, and unequal scan sizes could raise a broadcasting error. Reuse the nearest-neighbor distances for the stopping criterion.

Additional information

The regression checks a reordered identical scan and a shorter scan with known correspondences and nonzero residual. This changes association scoring independently of the transform accumulation in #1453. The regression failed on the original implementation and passed after the fix. Tested with Python 3.13.17 and the repository's pinned requirements. The modified module's tests and examples passed, Ruff passed, and the Sphinx HTML build succeeded.

CheckList

  • Added regression coverage for the defect.
  • Updated the example documentation.
  • CI results pending.

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant