Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

DCReg: Decoupled Characterization for Efficient Degenerate LiDAR Registration

Xiangcheng Hu, Xieyuanli Chen, Mingkai Jia, Jin Wu, Ping Tan, Steven L. Waslander

Abstract

LiDAR point cloud registration is fundamental to robotic perception and navigation. In geometrically degenerate environments (e.g., corridors), registration becomes ill-conditioned: certain motion directions are weakly constrained, causing unstable solutions and degraded accuracy. Existing detect-then-mitigate methods fail to reliably detect, physically interpret, and stabilize this ill-conditioning without corrupting the optimization. We introduce DCReg (Decoupled Characterization for Ill-conditioned Registration), establishing a detect-characterize-mitigate paradigm that systematically addresses ill-conditioned registration via three innovations. First, DCReg achieves reliable ill-conditioning detection by employing Schur complement decomposition on the Hessian matrix. This decouples the 6-DoF registration into 3-DoF clean rotational and translational subspaces, eliminating coupling effects that mask degeneracy in full-Hessian analyses. Second, within these subspaces, we develop interpretable characterization techniques resolving eigen-basis ambiguities via basis alignment. This establishes stable mappings between eigenspaces and physical motion directions, providing actionable insights on which motions lack constraints and to what extent. Third, leveraging this spectral information, we design a targeted mitigation via a structured preconditioner. Guided by MAP regularization, we implement eigenvalue clamping exclusively within the preconditioner rather than modifying the original problem. This preserves the least-squares objective and minimizer, enabling efficient optimization via Preconditioned Conjugate Gradient with a single interpretable parameter. Experiments demonstrate DCReg achieves 20-50% higher long-duration localization accuracy and 5-30x speedups (up to 116x) over degeneracy-aware baselines across diverse environments. Code: https://github.com/JokerJohn/DCReg

DCReg: Decoupled Characterization for Efficient Degenerate LiDAR Registration

Abstract

LiDAR point cloud registration is fundamental to robotic perception and navigation. In geometrically degenerate environments (e.g., corridors), registration becomes ill-conditioned: certain motion directions are weakly constrained, causing unstable solutions and degraded accuracy. Existing detect-then-mitigate methods fail to reliably detect, physically interpret, and stabilize this ill-conditioning without corrupting the optimization. We introduce DCReg (Decoupled Characterization for Ill-conditioned Registration), establishing a detect-characterize-mitigate paradigm that systematically addresses ill-conditioned registration via three innovations. First, DCReg achieves reliable ill-conditioning detection by employing Schur complement decomposition on the Hessian matrix. This decouples the 6-DoF registration into 3-DoF clean rotational and translational subspaces, eliminating coupling effects that mask degeneracy in full-Hessian analyses. Second, within these subspaces, we develop interpretable characterization techniques resolving eigen-basis ambiguities via basis alignment. This establishes stable mappings between eigenspaces and physical motion directions, providing actionable insights on which motions lack constraints and to what extent. Third, leveraging this spectral information, we design a targeted mitigation via a structured preconditioner. Guided by MAP regularization, we implement eigenvalue clamping exclusively within the preconditioner rather than modifying the original problem. This preserves the least-squares objective and minimizer, enabling efficient optimization via Preconditioned Conjugate Gradient with a single interpretable parameter. Experiments demonstrate DCReg achieves 20-50% higher long-duration localization accuracy and 5-30x speedups (up to 116x) over degeneracy-aware baselines across diverse environments. Code: https://github.com/JokerJohn/DCReg

Paper Structure

This paper contains 61 sections, 6 theorems, 78 equations, 19 figures, 9 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Motivation and Challenges
  3. Contributions
  4. Related Work
  5. Ill-Conditioning Detection
  6. Spectral Analysis Methods
  7. Adaptive Modeling Methods
  8. Ill-Conditioning Characterization
  9. Ill-Conditioning Mitigation
  10. Numerical Methods
  11. Robust Registration Approaches
  12. Research Gap
  13. Problem Formulation and Preliminaries
  14. Notations
  15. Nonlinear Point-to-Plane Registration
  16. ...and 46 more sections

Key Result

Theorem 1

For the Hessian $\bm{H}$, the Schur complements$\bm{S}_R$ and $\bm{S}_t$ satisfy: Analogous results hold for $\bm{S}_t$ when eliminating $\bm{\phi}$.

Figures (19)

  • Figure 1: Physical interpretation of Hessian eigendecomposition in ICP for a 3-DoF decoupled subspace (e.g., rotational block). The transformation between the canonical parameter basis and eigenspace reveals degenerate directions and their projections in the parameter space, enabling principled degeneracy characterization (\ref{['subsec:structural_ill_conditioning']}).
  • Figure 2: The error amplification effect in ill-conditioning. Visualization of how condition number determines the upper bound on relative parameter error given relative perturbations in the optimization space (\ref{['subsec:condition_define']}).
  • Figure 3: Visualization of rotation-translation coupling effect on ill-conditioning detection. (a) Traditional diagonal subspace decoupling ignores cross-coupling terms, which may hide existing ill-conditioning. (b) Schur complement decoupling properly isolates subspaces by accounting for coupling effects, enabling reliable degeneracy detection in the clean parameter subspace (\ref{['subsec:decoupling_theory']}).
  • Figure 4: Pipeline for quantitative ill-conditioning characterization. The three-stage process addresses eigenvector ambiguities in optimization subspaces: (1) Inner product matching resolves sign ambiguity and determines linear combinations in parameter space (\ref{['subsec:inner_product']}); (2) Maximum component analysis resolves ordering ambiguity to identify principal axis alignment (\ref{['subsec:max_component']}); (3) Gram-Schmidt orthogonalization resolves basis ambiguity, producing a stable orthonormal basis aligned with the parameter space for subsequent degeneracy mitigation (\ref{['subsec:gram_schmidt']}).
  • Figure 5: Algorithmic principles for spectral shaping. Visualization of how different methods shape the eigenvalue spectrum to mitigate degeneracy: traditional regularization directly modifies the Hessian, whereas our method uses targeted eigenvalue clamping only in a structure-aware preconditioner.
  • ...and 14 more figures

Theorems & Definitions (10)

  • Theorem 1: Elimination-Curvature Equivalence
  • Proposition 1: Projection Representation
  • Proposition 2: Scale- and Orthogonal-Basis Invariance
  • Theorem 2: Spectral Bounds for Schur Conditioning
  • Remark 1
  • Remark 2
  • Remark 3
  • Theorem 3: Equivalent Reconstruction on Decoupled Subspace
  • Theorem 4: MAP Interpretation of Eigenvalue Clamping
  • Remark 4