Research Analyzer
← Back ICRA 2026

Cubemap-Based LiDAR-Inertial Odometry with Intensity Assistance

Yang Liu, Kazushige YAMAMOTO, Atsushi Matsui, Saburo Takahashi, Toshihisa Abe

PDF

AI summary

Key figure (auto-extracted from paper)
CUBE-LIO significantly improves LiDAR odometry accuracy and robustness in geometrically degenerate environments by efficiently leveraging cubemap-projected intensity data for direct photometric optimization.
LiDAR odometry photometric optimization cubemap projection intensity-assisted SLAM degenerate environments tightly coupled LIO

Problem

LiDAR-only odometry degrades in geometrically degenerate environments like tunnels or open spaces. Existing intensity-assisted methods rely on equirectangular projection, which introduces severe polar distortion, high computational cost, and poor compatibility with solid-state LiDARs.

Approach

The framework projects LiDAR intensity onto a cubemap to eliminate polar distortion and accelerate computation, then applies a semi-dense intensity gradient magnitude selection strategy to jointly optimize photometric and geometric constraints in a tightly coupled LiDAR-inertial pipeline.

Key results

  • Cubemap projection eliminates polar distortion and reduces computational overhead compared to equirectangular mapping
  • Semi-dense IGM feature selection enhances resilience to intensity noise and measurement variations
  • Achieves state-of-the-art accuracy and real-time performance across multiple public benchmarks
  • Sensor-agnostic architecture supports both spinning and solid-state LiDARs

Why it matters

Enables reliable, low-cost navigation for robotics and autonomous systems in challenging environments without requiring additional cameras or complex sensor calibration.

Abstract

We present CUBE-LIO, a LiDAR-inertial odom- etry framework that leverages direct photometric constraints from LiDAR intensity to improve robustness in geometrically degenerate environments. At its core is an efficient cubemap projection that maps LiDAR intensity onto six cube faces, eliminating pole singularities and severe polar distortion. This yields a more uniform spatial sampling while avoiding the costly trigonometric operations typical of equirectangular mappings. Building on this representation, we introduce a semi-dense feature selection and direct optimization strategy based on intensity gradient magnitude. This strategy improves resilience to intensity noise and variations induced by range and incidence angle. Photometric constraints are jointly optimized with geo- metric measurements in a tightly coupled LIO pipeline. CUBE- LIO is sensor-agnostic and supports both spinning and solid- state LiDARs. Experiments on multiple public benchmarks demonstrate state-of-the-art accuracy and real-time perfor- mance, with particularly pronounced gains in scenes where the geometric structure is sparse or weak.

Index terms

SLAM Localization Mapping

Related papers