LiDAR Artifact Removal Techniques

Executive Summary

LiDAR artifact removal is a layered stack, not a single filter. It covers classical outlier removal, weather denoising, sensor health diagnostics, ghost and multipath suppression, dynamic-object masking, and static-map cleaning. The correct safety question is not "which method removes the most points?" but "which method removes false measurements without hiding real hazards or weakening localization observability?"

For airside autonomous vehicles, the broad removal layer should include:

• Classical filters: SOR, ROR, DROR, DSOR, LIOR, DDIOR, D-LIOR, IDSOR, DVIOR, SDOR, LIDSOR.
• Learned weather removal where supported: LIORNet and related denoisers, validated against classical baselines.
• Sensor artifact handling: ghost, multipath, retroreflector blooming, sun/receiver saturation, blockage, and dust.
• Dynamic map cleaning: ERASOR, Removert, MapCleaner, ERASOR++, 4dNDF, and MOS-style evaluation.
• Safety validation: raw-vs-filtered evidence, target-domain labels, SOTIF argumentation, and ODD degradation rules.

Repo Cross-Links

Topic	Link	Role
Classical filters	Classical LiDAR Outlier Removal	Baseline and deterministic weather filters.
Weather artifacts	LiDAR Weather Artifact Removal	Snow, rain, fog, dust, spray, mist, and wet-surface artifacts.
Sensor ghosts	LiDAR Ghost and Multipath Artifacts	Reflective surfaces, multipath, bloom, and saturation.
Safety validation	LiDAR Artifact Removal Validation	Evidence plan and airside validation gates.
Dynamic map cleaning	LiDAR Map Cleaning and Dynamic Removal	Static map construction and dynamic clutter removal.
ERASOR	ERASOR	Pseudo-occupancy dynamic object removal.
Removert	Removert	Remove-then-revert static map cleaning.

Artifact Taxonomy

Artifact class	Examples	Primary symptom	Best first response
Isolated statistical outliers	Random invalid returns, edge speckle	Sparse points away from surfaces	SOR/ROR with diagnostics.
Weather particles	Snow, rain, dust, spray	Near-field false points, attenuation, low persistence	DROR/DSOR/SDOR/LIOR variants plus temporal checks.
Aerosol volume	Fog, steam, de-icing mist	Range collapse, volumetric backscatter	ODD degradation and radar-primary mode.
Reflective ghosts	Glass, wet ground, polished aircraft skin	Mirrored or behind-surface points	Reflective-surface reasoning, waveform/multi-return, map consistency.
Saturation/bloom	Retroreflective signs, vests, markings, direct sun	Inflated target, angular dropout, high intensity	Per-sector intensity and receiver health checks.
Sensor blockage	Dirt, glycol film, ice, bug splat	No-return sectors, depth-image holes	Autoware-style blockage diagnostics and cleaning.
Dynamic objects	Aircraft, tugs, carts, buses, people	Trails in accumulated maps, scan-to-map residuals	MOS, ERASOR, Removert, MapCleaner, 4dNDF.
Map staleness	Construction, moved stand equipment	Persistent disagreement with map	Change detection and map lifecycle workflow.

Technique Taxonomy

Layer	Techniques	Output	Safety role
Input sanity	NaN removal, crop boxes, min/max range, vehicle-body masking	Valid raw cloud	Prevents impossible data from entering the stack.
Classical filtering	SOR, ROR, DROR, DSOR, LIOR, DDIOR, D-LIOR, IDSOR, DVIOR, SDOR, LIDSOR	Clean cloud plus removed cloud	Explainable baseline and weather control.
Learned denoising	LIORNet, WeatherNet-style, 4D temporal denoisers	Per-point noise probability	Better complex weather handling, but needs validation and uncertainty.
Sensor health	Blockage, dust, sector coverage, intensity drift, max range	Degradation state	Drives cleaning, speed limiting, and ODD enforcement.
Ghost handling	Reflective plane detection, full-waveform ghost removal, PCL ShadowPoints	Ghost mask	Prevents false objects and false map structure.
Dynamic masks	LiDAR-MOS, 4DMOS, HeLiMOS-style MOS	Moving/static labels	Protects localization and maps from moving actors.
Static map cleaning	ERASOR, Removert, MapCleaner, ERASOR++, 4dNDF	Static map plus rejected dynamic layer	Creates long-term localization maps.

Deployment Decision Rules

Situation	Recommended behavior	Avoid
Clear weather, healthy sensor	Light SOR/ROR and artifact diagnostics.	Aggressive weather filters that reduce useful map geometry.
Light snow or rain	Range-aware DROR/DSOR/SDOR plus intensity-aware checks.	Fixed global ROR/SOR thresholds as the only protection.
Heavy snow, fog, dust, or de-icing mist	Reduced speed, radar-primary perception, sensor health alerts.	Claiming LiDAR is clean because a denoiser returned a dense-looking cloud.
Wet apron or reflective terminal area	Ground-model and multipath diagnostics, camera/radar agreement.	Treating below-ground points as real obstacles or deleting all low returns.
Retroreflective apron markings	Intensity saturation checks and known-object geometry bounds.	Letting bloom enlarge object boxes or map features.
Static map build	Combine dynamic masks, ERASOR/Removert/MapCleaner, multi-session consensus.	Building a localization map from a single busy shift.
Runtime localization	Downweight dynamic/artifact points but monitor static inlier count.	Removing so many points that scan matching becomes unobservable.

Failure Modes

• False deletion of small real obstacles under aggressive weather filtering.
• False retention of coherent artifacts such as spray sheets, glass ghosts, or wet-surface mirrors.
• Intensity threshold transfer failure across sensor vendors or sensor covers.
• Removing dynamic objects from maps but accidentally eroding static ground, poles, gate equipment, and aircraft stand features.
• Cleaner-induced domain shift for detectors trained on raw clouds.
• ODD monitor blind spot: the filter removes many points but no one notices the sensor is no longer adequate for the vehicle speed.
• Map lifecycle error: temporary parked aircraft or GSE is promoted into the long-term static map.

Airside-Specific Validation Guidance

Airside validation needs target-domain clips and point labels. Include:

• Weather: heavy rain, snowfall, fog, dust, road spray, de-icing mist, steam, and glycol cover contamination.
• Reflective conditions: wet concrete, painted stand markings, retroreflective signs, cones, high-vis clothing, aircraft fuselage, terminal glass.
• Dynamic scenes: moving and parked aircraft, tugs, buses, belt loaders, dollies, fuel trucks, chocks, cones, and ground crew.
• Localization stress: open apron, repeated gates, terminal-edge multipath, wet night operations, and GNSS-challenged areas.
• Map lifecycle: same stand across shifts, aircraft present/absent, construction, temporary barriers, and seasonal snow banks.

Minimum evidence package:

• Raw, filtered, and removed point clouds.
• Per-artifact confusion matrix.
• Detector and tracker before/after metrics.
• Localization inlier, residual, and degeneracy metrics.
• Static map ghost rate and static preservation rate.
• ODD transition logs showing speed reduction, radar-primary mode, cleaning, or controlled stop.

Practical Recommendation

Build removal in layers:

1. Keep raw data and diagnostics.
2. Apply conservative classical filters.
3. Add weather-specific filters with explicit activation conditions.
4. Detect sensor blockage and saturation independently from denoising.
5. Use radar/camera/thermal confirmation for safety-critical deletion.
6. Clean maps offline with multi-session evidence.
7. Validate against airside-labeled artifacts before using removal as safety evidence.

Sources

• Open3D outlier removal: https://www.open3d.org/docs/latest/tutorial/Advanced/pointcloud_outlier_removal.html
• PCL filters: https://pointclouds.org/documentation/group__filters.html
• DSOR and WADS: https://arxiv.org/abs/2109.07078
• DDIOR: https://www.mdpi.com/2072-4292/14/6/1468
• DVIOR: https://www.mdpi.com/2079-9292/14/18/3662
• IDSOR: https://arxiv.org/abs/2602.05876
• SDOR: https://www.nature.com/articles/s41598-026-38674-6
• LIORNet: https://arxiv.org/abs/2603.19936
• ERASOR: https://arxiv.org/abs/2103.04316
• ERASOR repository: https://github.com/LimHyungTae/ERASOR
• Removert repository: https://github.com/gisbi-kim/removert
• 4dNDF: https://arxiv.org/abs/2405.03388
• 4dNDF repository: https://github.com/PRBonn/4dNDF
• HeLiMOS dataset: https://sites.google.com/view/helimos/dataset
• HeLiMOS toolbox: https://github.com/url-kaist/HeLiMOS-PointCloud-Toolbox
• Autoware blockage diagnostics: https://autowarefoundation.github.io/autoware_universe/pr-10077/sensing/autoware_pointcloud_preprocessor/docs/blockage-diag/
• ISO 21448 SOTIF: https://www.iso.org/standard/77490.html