Research Analyzer
← Back ICRA 2026

Onion-LO++: An Adaptive and Degradation Resistant Continuous-Time LiDAR Odometry

Xiaolong Cheng, Ye Sun, Keke Geng, Tianxiao Ma, Zhichao Liu

PDF

AI summary

Key figure (auto-extracted from paper)
Onion-LO++ enables robust, high-accuracy LiDAR-only odometry in extreme, degenerate, and high-motion scenarios by combining adaptive point cloud segmentation with a continuous-time trajectory model.
LiDAR odometry continuous-time trajectory point cloud segmentation degenerate environments high-motion scenarios adaptive optimization

Problem

Existing LiDAR-only odometry methods struggle with geometric degradation in planar environments and instability during high-speed motions, while multi-sensor fusion systems introduce calibration, synchronization, and computational overhead.

Approach

The method uses a coarse-to-fine segmentation strategy to extract intensity and weak edge features from planar regions, dynamically adjusts downsampling based on scene degeneracy, and integrates a continuous-time trajectory model with an adaptive onion factor to optimize parameters in real-time.

Key results

  • Superior accuracy and robustness on five challenging public datasets
  • Reliable operation in narrow spaces, degenerate scenes, and high-speed motion
  • Real-time adaptive optimization via a dynamic onion factor
  • Elimination of IMU calibration and synchronization requirements

Why it matters

Provides a robust, calibration-free alternative to sensor-fusion odometry for extreme robotics applications like UAVs and handheld mapping.

Abstract

In an era dominated by multi-sensor fusion, this paper explores the operational limits of LiDAR-only odometry. We introduce Onion-LO++, which is designed to overcome two practical limitations of Onion-LO: poor performance in geometrically degenerate environments and instability under high-motion conditions. In order to mitigate point cloud degradation, we propose a coarse-to-fine point cloud segmentation approach that extracts intensity and weak corner features from planar regions, while dynamically adjusting the downsampling rate based on the proportion of planar points to maximize geometric constraints. To handle high-motion scenarios, we integrate a continuous-time trajectory model into the backend optimization and introduce an adaptive onion factor that adjusts optimization parameters in real time. Extensive experiments on five challenging public datasets demonstrate that Onion-LO++ outperforms state-of-the-art methods and operates reliably across narrow spaces, degenerate scenes, high-speed motion, and high-altitude aerial mapping. We open-source the code on GitHub.1

Index terms

SLAM Field Robots

Related papers