When running ICP(for example with SLAM) it's possible your next position does not have a point cloud which matches up with the prior point cloud due to resolution limits. How is this dealt with it? You could artificially add points to the prior point cloud(ex, linear interpolation though I'm not sure how that works in 3d).


1 Answer 1


You probably mean ICP, not IPC.

The most typical way is using plane features. Plane information could be extracted by voxelisation and NDT.

Plane-to-point or plane-to-plane is enough for the registration of sparse points. You don't need to do the interpolation at all.

  • $\begingroup$ Thanks, how would that work though? You would need to extract the planes using something like RANSAC via parameter estimation and then what? $\endgroup$ May 20, 2022 at 14:02
  • $\begingroup$ Try to search plane-to-point ICP and NDT ICP. $\endgroup$ May 22, 2022 at 11:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.