Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Loop Closure via Maximal Cliques in 3D LiDAR-Based SLAM

Javier Laserna, Saurabh Gupta, Oscar Martinez Mozos, Cyrill Stachniss, Pablo San Segundo

TL;DR

A novel deterministic algorithm, CliReg, is introduced for loop closure validation that replaces RANSAC verification with a maximal clique search over a compatibility graph of feature correspondences, which avoids random sampling and increases robustness in the presence of noise and outliers.

Abstract

Reliable loop closure detection remains a critical challenge in 3D LiDAR-based SLAM, especially under sensor noise, environmental ambiguity, and viewpoint variation conditions. RANSAC is often used in the context of loop closures for geometric model fitting in the presence of outliers. However, this approach may fail, leading to map inconsistency. We introduce a novel deterministic algorithm, CliReg, for loop closure validation that replaces RANSAC verification with a maximal clique search over a compatibility graph of feature correspondences. This formulation avoids random sampling and increases robustness in the presence of noise and outliers. We integrated our approach into a real- time pipeline employing binary 3D descriptors and a Hamming distance embedding binary search tree-based matching. We evaluated it on multiple real-world datasets featuring diverse LiDAR sensors. The results demonstrate that our proposed technique consistently achieves a lower pose error and more reliable loop closures than RANSAC, especially in sparse or ambiguous conditions. Additional experiments on 2D projection-based maps confirm its generality across spatial domains, making our approach a robust and efficient alternative for loop closure detection.

Loop Closure via Maximal Cliques in 3D LiDAR-Based SLAM

TL;DR

A novel deterministic algorithm, CliReg, is introduced for loop closure validation that replaces RANSAC verification with a maximal clique search over a compatibility graph of feature correspondences, which avoids random sampling and increases robustness in the presence of noise and outliers.

Abstract

Reliable loop closure detection remains a critical challenge in 3D LiDAR-based SLAM, especially under sensor noise, environmental ambiguity, and viewpoint variation conditions. RANSAC is often used in the context of loop closures for geometric model fitting in the presence of outliers. However, this approach may fail, leading to map inconsistency. We introduce a novel deterministic algorithm, CliReg, for loop closure validation that replaces RANSAC verification with a maximal clique search over a compatibility graph of feature correspondences. This formulation avoids random sampling and increases robustness in the presence of noise and outliers. We integrated our approach into a real- time pipeline employing binary 3D descriptors and a Hamming distance embedding binary search tree-based matching. We evaluated it on multiple real-world datasets featuring diverse LiDAR sensors. The results demonstrate that our proposed technique consistently achieves a lower pose error and more reliable loop closures than RANSAC, especially in sparse or ambiguous conditions. Additional experiments on 2D projection-based maps confirm its generality across spatial domains, making our approach a robust and efficient alternative for loop closure detection.
Paper Structure (16 sections, 3 equations, 3 figures, 2 tables)

This paper contains 16 sections, 3 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. CliReg-Based Loop Closures
  4. Preliminaries
  5. Clique-Based Registration
  6. Feature Extraction and Encoding
  7. Descriptor Database and Correspondence Graph Construction
  8. Experimental Evaluation
  9. Metrics and Experimental Setup
  10. 2D and 3D Pipeline Description
  11. Evaluation Results
  12. Runtime Performance and Variability
  13. Case Study: Detailed Failure and Success Example
  14. Failure Modes in 2D Scenarios
  15. Summary
  16. ...and 1 more sections

Figures (3)

  • Figure 1: Successful loop closure association between two local LiDAR submaps in a complex urban environment. The red and green points represent keypoints from the query and reference maps, respectively. The left image shows the result of our approach, while the right image shows the result of a RANSAC-based method.
  • Figure 2: An overview of our loop closure detection and validation pipeline. It consists of three stages: (1) Feature Extraction and Encoding, (2) Correspondence Graph generation, and (3) Pose Estimation via maximal clique search. This integration enables efficient and robust loop closure detection across 2D and 3D feature representations.
  • Figure 3: Example of a correspondence graph $G$ with nodes representing feature matches and edges denoting geometric consistency. The highlighted 5-clique forms the basis for $\mathbb{SE}(3)$ transformation.