CHANGELOG

Changelog for package mola_sm_loop_closure

1.2.0 (2026-05-11)

• f2f pipeline: switch to multi-objective as default strategy

• Merge pull request #15 from MOLAorg/simplify-ci CI: simplify clang-format helpers and use ros: docker images for stable builds

• fix: wrap find -iname predicates in parentheses to scope them to listed dirs Without grouping, the -o operators apply at top level and -print0 only attaches to the last alternative, so files outside the intended directories can be matched.

• CI: simplify clang-format helpers and use ros: docker images for stable builds - Replace complex Python-based clang_git_format with a simple scripts/formatter.sh supporting –check - Standardize formatter to scripts/formatter.sh (moved from root formatter.sh) - Simplify check-clang-format.yml to just apt-install clang-format-14 and run the script - Use ros:humble / ros:jazzy pre-built images for stable CI builds (faster, no setup-ros needed)

• fix:as libflann as build_depend as needed for kiss-matcher

• CI: sensible job names

• bump min req cmake version to 3.22

• feat: live gui, show GNSS data points

• update package.xml build deps

• Merge pull request #14 from MOLAorg/improve-lc-hypothesis-coverage Improve lc hypothesis coverage

• fix: deduplicate candidates by block-pair ID before selection in find_loop_candidates Multiple (i,j) frame pairs can round to the same (frameGroup_i, frameGroup_j) block and all pass the alreadyChecked filter (which only holds pairs from previous rounds). All of them survived into the scored candidate list, so max_lc_candidates were returned but the evaluation loop in process() silently skipped all but the first occurrence of each block pair – causing far fewer ICP evaluations than expected. Fix: after scoring (step 1) and before selection (step 2), deduplicate each candidate container by block-pair ID, keeping only the highest-scored entry per block. This is applied to both the flat candidates vector (PROXIMITY_ONLY, MULTI_OBJECTIVE) and each distance bin (DISTANCE_STRATIFIED), so max_lc_candidates now matches the number of candidates that will actually be evaluated. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

• fix: MULTI_OBJECTIVE coverage term now uses spatial midpoints instead of sin formula The old coverage score used abs(sin(pi * topologicalMidpoint / totalFrames)), which only rewards candidates near the center of the frame index range and ignores actual map geometry. This caused clustering in the same physical region on non-linear trajectories (U-shapes, figure-8s, long straights). New approach: - Add LoopCandidate::spatialMidpoint ((pose_i + pose_j) / 2, computed once per candidate in find_loop_candidates()). - Track selectedMidpoints alongside selectedDistances in the greedy selection loop. - Coverage score penalizes candidates whose spatial midpoint is within ~20 m of an already-selected midpoint, forcing the selected set to cover different map areas regardless of trajectory shape. - Refactor score_multi_objective() to accept a MultiObjectiveArgs struct instead of a long flat parameter list, making call sites self-documenting. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

• fix: MULTI_OBJECTIVE diversity scoring is now actually applied during selection selectedDistances was declared but never populated, so score_multi_objective() always computed diversity against an empty set (score = 1.0 always). The fix replaces the simple sort+truncate step 2 path for MULTI_OBJECTIVE with a greedy selection loop: pick the best candidate, record its distance, re-score the rest with the updated selectedDistances, repeat. This makes the diversity term a real participant in selection instead of a no-op. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

• fix: live preview GNC stats no longer reset to zero at each LC round start liveStats was zero-initialized at the top of every lcRound loop, causing the 3Dscene caption to show “0 inliers, 0 outliers” until the first GNC optimization ran in that round. Carry lastGncInliers / lastGncOutliers forward so the caption always reflects the most recent optimizer result. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

• feat: show better inliers/outliers stats

• feat: manual LC now has an optional bypass for inliers

• Merge pull request #13 from MOLAorg/feat/live-gui-3dscene-file feat: show ‘live’ progress via 3Dscene file updates

• feat: show ‘live’ progress via 3Dscene file updates

• add new mp2p_icp flags: force_final_pairings_for_quality

• Add Codecov badge to README

• Merge pull request #12 from MOLAorg/feat/f2f-integrate-kiss-matcher feat: Incorporate KISS-Matcher as optional seeder for initial ICP poses

• fix: CI code coverage

• chore: add more INFO traces

• Fixes: cmake gating, don’t expose kiss-matcher headers, manual LC should be inlier

• feat: Incorporate KISS-Matcher as optional seeder for initial ICP poses

• FIX: crashes due to missing noise model and different gtsam symbols

• feat: add gnss filter by max uncertainty

• Merge pull request #11 from MOLAorg/refactor-common-code refactor: unify common code between the two LC algorithms

• CI: Fix for new ROS2 rolling, add codecov upload

• refactor: unify common code between the two LC algorithms

• Contributors: Jose Luis Blanco-Claraco

1.1.0 (2026-04-29)

• Merge pull request #10 from MOLAorg/refactor/lc-common-helpers refactor: extract lc_common helpers; port planar-world annealing + GNC to SM

• refactor: replace two-pass LM with single GNC pass in SimplemapLoopClosure

• refactor: extract lc_common helpers; port planar-world annealing + GNC to SM

• Merge pull request #9 from MOLAorg/feat/planar-world-in-f2f feat: optional f2f planar world soft-constraints

• feat: optional f2f planar world soft-constraints

• Merge pull request #8 from MOLAorg/feat/manual-lc Support for optional manual hints for LC

• Support for optional manual hints for LC

• Merge pull request #7 from MOLAorg/feat/more-flexible-icp-sigmas-and-logging Expose more env vars

• Expose more env vars

• f2f pipeline file: add env var MOLA_DESKEW_IGNORE_ACCELEROMETER

• Contributors: Jose Luis Blanco-Claraco

1.0.0 (2026-04-14)

• package.xml: add missing test-dep

• add unit tests

• cli: use output directory as default for debug output files, not input

• Install pipelines so they are accessible under ‘share’

• Make the package discoverage by ament

• Merge pull request #6 from MOLAorg/save-ram F2F algorithm: save ram in offline runs

• More memory efficient loop

• Bump minimum cmake version to 3.7

• Reuse mp2p_icp::update_velocity_buffer_from_obs() from mp2p_icp Removes duplicated code in this repo now that mp2p_icp>=2.5.0 is available in all distributions

• Fix build against mp2p_icp <2.6.0

• icp debug log files: save only good edges

• F2F algorithm: save ram in offline runs

• sm2mm pipeline yaml: expose more params via env vars

• Expose more parameters for F2F algorithm

• Debug feature: enable saving 3Dscene files per optimization round

• Use Graduated Non-Convexity (GNC) optimizer for superior outlier rejection

• FIX: Reverted logic in formula for adaptive threshold

• Add formal CLA

• Merge pull request #5 from MOLAorg/feat/smart-ram-lazy-unload Lazy unload keyframe clouds to keep RAM usage bounded

• Contributors: Jose Luis Blanco-Claraco

0.2.0 (2026-03-03)

• Fix build against different gtsam versions

• Add optional generation of .3Dscene files with loop-closure visualizations

• Tolerate missing external files without throwing

• Better logging and expose more env var params

• Add GNSS uncertainty multiplier

• Use debug files prefix

• Add more frame-to-frame LC selection algorithms

• Merge pull request #4 from MOLAorg/feat/f2f-fixes Add more parameters to f2f method

• Add new params for f2f method

• pipeline file: expose more params and fix reversed sigma logic

• Fix: prior should be weak for GNSS to transform to ENU frame

• fix copyright headers

• Merge pull request #3 from MOLAorg/feat/new-f2f-algo Add alternative frame-to-frame LC algorithm

• Provide virtual LoopClosureInterface to select the algorithm from a CLI argument

• Add alternative frame-to-frame LC algorithm

• Merge pull request #2 from MOLAorg/feat/add-ci-and-badges Add CI workflows and clang-format linting infrastructure

• package.xml: add FILE tag

• Merge pull request #1 from MOLAorg/feature/refactor-pkg-mola-gtsam-factors Progress moving code to new package mola_gtsam_factors

• Update dependencies after refactorization into mola_georeferencing

• Progress moving code to new package mola_gtsam_factors

• Protect against invalid GPS cov entries

• Add param gnss_minimum_uncertainty_xyz; dump trajectories as TUM files.

• simplify: remove obsolete parameters not used anymore here

• Remove unused parameter (GPS observations are automatically detected anyway)

• Fix: ensure minimum uncertainty in edges

• Fix deskewing generating sm2mm for local maps

• Implement missing loading of new LocalVelocityBuffer while doing submaps “sm2mm”

• Don’t use absolute paths for debug icplog files

• FIX: Detect the error case of no GNSS observations in sm-georeference-cli

• Add gicp pipeline

• bump minimum cmake version to 3.5

• add .clang-tidy rules and better vscode integration

• Expose more env vars in the pipeline

• robust kernels; optional horizontality

• progress with submaps

• implement submaps

• use new mrpt enu covariance field

• done georeferencing cli

• progress

• progress cli app interface

• initial structure

• Initial commit

• Contributors: Jose Luis Blanco Claraco, Jose Luis Blanco-Claraco