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

High-quality maps are fundamental for robotics tasks such as navigation and planning. Although modern graph-based LiDAR SLAM systems achieve good trajectory accuracies, a low trajectory error alone does not guarantee geometrically consistent maps, particularly at revisit locations where missed loop closures and residual drift can produce local misalignments. In this work, we address the problem of jointly improving global trajectory estimation and local map quality in 3D LiDAR SLAM. We first propose a framework to efficiently estimate geometry-dependent information matrices for ICP, enabling principled weighting of odometry constraints in a pose graph. We then introduce a hierarchical loop-closure module that decouples place recognition from geometric registration, together with a retroactive loop-closure module that exploits the optimized pose graph to recover missed loop closures. We also propose an evaluation protocol to measure map consistency at revisit locations. We evaluate our SLAM system on several datasets against state-of-the-art LiDAR SLAM systems. Experimental results demonstrate global trajectory accuracies on par with or better than existing methods while consistently improving local geometric map consistency at revisit locations. These results suggest that coupling uncertainty-aware odometry with geometry-guided loop-closure refinement leads to more accurate trajectories and higher-quality maps.

False LiDAR loop closures can distort pose graphs in repetitive environments, where one query may match several distant places. Genuine revisits, unlike many perceptual aliases, tend to remain consistent across neighboring keyframes. We present Query-Calibrated Segmental Admission (QCSA), a loop-candidate selection method between descriptor retrieval and point-cloud registration. QCSA maps descriptor scores to evidence that distinguishes true loops from hard negatives, accumulates it over short keyframe segments, and admits only the best segment above a threshold fitted to the strongest false segment for each calibration query. The admitted segment forms two local submaps for one Generalized Iterative Closest Point (GICP) registration. QCSA requires pairwise descriptor scores and separate calibration for each descriptor and setting. Across seven descriptors on the SNU Library Dataset (SNULib), QCSA lowers average trajectory error with 53.8% fewer registration attempts than Top-1. With common local submaps and downstream processing on NCLT, it also reduces registration attempts, constraints exceeding the pose-error tolerance, and trajectory error relative to Top-1. HeLiPR Bridge further evaluates repeated outdoor geometry. Our implementation and dataset will be released at: https://github.com/wanderingcar/snu_library_dataset.

Loop closure is essential to reduce drift and build globally consistent maps in large-scale environments. However, reliable loop closure with only geometric information from, e.g., a LiDAR sensor, remains challenging due to the difficulty of constructing discriminative geometric features. We present PinNet, a neural network that produces local geometric descriptors from point clouds for place recognition and scanto-scan registration. PinNet incorporates a neural network that generates keypoints and their corresponding descriptors, together with a plane-based geometric self-attention module that models inter-keypoint spatial relationships to enhance descriptor discriminability for loop-closure detection and point-cloud registration. The approach is comprehensively evaluated on multiple datasets collected with different LiDAR sensors. Experimental results demonstrate strong place-recognition performance, precise relative pose estimation, and successful single-shot localization in different environments.