Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix ICP sample indices after initialization #183

Open
keenan-burnett opened this issue May 11, 2023 · 0 comments
Open

Fix ICP sample indices after initialization #183

keenan-burnett opened this issue May 11, 2023 · 0 comments
Assignees
Labels
enhancement New feature or request

Comments

@keenan-burnett
Copy link
Contributor

Currently, we resample the filtered_samples_inds at each iteration of ICP:

for (size_t i = 0; i < sample_inds.size(); i++) {
if (nn_dists[i] < max_pair_d2) {
// Check planar distance (only after a few steps for initial alignment)
auto diff = aligned_points[sample_inds[i].first].getVector3fMap() -
point_map[sample_inds[i].second].getVector3fMap();
float planar_dist = std::abs(
diff.dot(point_map[sample_inds[i].second].getNormalVector3fMap()));
if (step < first_steps || planar_dist < max_planar_d) {
filtered_sample_inds.push_back(sample_inds[i]);
}
}
}
timer[2]->stop();
/// point to plane optimization

This essentially "moves" the minimum of the cost function at each iteration of ICP.
If we fixed filtered_samples_inds after the initialization steps, it might allow ICP to converge faster.

Currently, it often takes 20-40 iterations of ICP to converge which seems like too much.

@keenan-burnett keenan-burnett added the enhancement New feature or request label May 11, 2023
@keenan-burnett keenan-burnett self-assigned this May 11, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request
Projects
None yet
Development

No branches or pull requests

1 participant