LiMo-Calib: On-Site Fast LiDAR-Motor Calibration for Quadruped Robot-Based Panoramic 3D Sensing System

Jianping Li*, Zhongyuan Liu*, Xinhang Xu, Jinxin Liu · 2025

Read LiMo-Calib: On-Site Fast LiDAR-Motor Calibration for Quadruped Robot-Based Panoramic 3D Sensing System by Jianping Li*, Zhongyuan Liu*, Xinhang Xu, Jinxin Liu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-12655.txt.

Introduction

Jianping Li and co-authors calibrate a spinning laser scanner on a four-legged robot so it can map in all directions, in work from 2025. A single fixed scanner leaves blind spots, a real limit on a payload-tight robot. A motor sweeps the scanner to widen the view.

Walking shakes the rig, so the scanner-to-motor alignment drifts and maps smear. Older fixes need targets or heavy feature work, unfit for the field. Their method reads plain geometry from raw scans, picks features by a spread rule to converge fast, and reweights patches by how well a flat fit holds. The team mounts the system on a legged robot and checks it live.

Catalog record

CREATORJianping Li*, Zhongyuan Liu*, Xinhang Xu, Jinxin Liu
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE35 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-12655

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

LiMo-Calib: On-Site Fast LiDAR-Motor Calibration for Quadruped Robot-Based Panoramic 3D Sensing System Jianping Li*, Zhongyuan Liu*, Xinhang Xu, Jinxin Liu Abstract Conventional single LiDAR systems are inherently constrained by their limited field of view (FoV), leading to blind spots and incomplete environmental awareness, particularly on robotic platforms with strict payload limitations. Integrating a motorized LiDAR offers a practical solution by significantly expanding the sensor’s FoV and enabling adaptive panoramic 3D sensing. However, the high-frequency vibrations of the quadruped robot introduce calibration challenges, causing variations in the LiDAR-motor transformation that degrade sensing accuracy. Existing calibration methods that use artificial targets or dense feature extraction lack feasibility for on-site applications and real-time implementation. To overcome these limitations, we propose LiMo-Calib, an efficient on-site calibration method that eliminates the need for external targets by leveraging geometric features directly from raw LiDAR scans. LiMo-Calib optimizes feature selection based on normal distribution to accelerate convergence while maintaining accuracy and incorporates a reweighting mechanism that evaluates local plane fitting quality to enhance robustness. We integrate and validate the proposed method on a motorized LiDAR system mounted on a quadruped robot, demonstrating significant improvements in calibration efficiency and 3D sensing accuracy, making LiMo-Calib well-suited for real-world robotic applications. We further demonstrate the accuracy improvements of the LIO on the panoramic 3D sensing system using the calibrated parameters. The code will be available at: https://github.com/kafeiyin00/LiMo-Calib . † † footnotetext: This work was supported by China-Singapore International Joint Research Institute (CSIJRI) Development Plan for the Technology Application Center (TAC) and National Research Foundation, Singapore, under its Medium-Sized Center for Advanced Robotics Technology Innovation. This work was partially supported by open project funding of Key Laboratory of Intelligent Control and Optimization for Industrial Equipment of Ministry of Education under Grant LICO2023YB01. J. Li, Z. Liu, X. Xu, J. Liu, S. Yuan, and L. Xie are with China-Singapore International Joint Research Institute (CSIJRI) and School of Electrical and Electronic Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore. X. Fang is with School of Control Science and Engineering, Dalian University of Technology, Dalian 116024, China (E-mail: [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected]) (Jianping Li and Zhongyuan Liu contribute equally to this work.) I Introduction In recent years, robotic platforms, particularly quadruped robots, have seen significant advancements, making them increasingly valuable for applications such as autonomous inspection, search and rescue, and exploration of complex environments [ 1 , 2 ] . However, conventional LiDAR systems are inherently constrained by their limited field of view (FoV), leading to blind spots and incomplete environmental awareness [ 3 ] . Given the weight and size constraints of quadruped robots, mounting multiple LiDAR units to enhance FoV is often impractical [ 4 , 5 , 6 , 7 , 8 ] . A more efficient solution is integrating a motorized LiDAR [ 9 ] onto the quadruped robot platform for a panoramic 3D sensing system as shown in Fig. 1 . By leveraging motor control, the motorized LiDAR substantially expands the sensor’s FoV, enabling broader spatial coverage and adaptive scanning to focus on critical features. Integrating a motorized LiDAR with a quadruped robot introduces significant calibration issues due to the high-frequency vibrations inherent in the robot’s movement [ 10 ] . These vibrations and oscillations cause fluctuations in the relative transformation between the LiDAR and the motor, leading to potential misalignment. If not compensated properly on site, these variations can severely degrade the accuracy of 3D sensing, compromising the reliability of the system. Figure 1: Coordinates and mechanical design of the proposed motorized LiDAR system on a quadruped robot. The calibration of the LiDAR system has been long studied in the field of robotics [ 11 , 12 ] and photogrammetry [ 13 , 14 ] . Artificial targets are usually used for calibration of the LiDAR system [ 15 ] . But the artificial targets may not be available for the on-site calibration applications. Despite the artificial targets distributed in the calibration filed, planar features extracted from the environment are the most commonly used primitives for automatically constructing the calibration functions [ 11 , 13 , 16 ] . Nevertheless, the large number of extracted planar features significantly increases computational complexity, leading to long processing times that are impractical for edge computing units on mobile robots. Existing LiDAR-motor calibration methods inherently involve a trade-off between the number of extracted primitives and calibration accuracy. A higher number of primitives can enhance accuracy by providing more constraints, but it also leads to increased computational demands and longer processing times [ 17 , 13 , 18 ] . Conversely, reducing the number of primitives improves efficiency but may compromise calibration precision. Achieving an optimal balance between these factors remains a key challenge in developing accurate and efficient on-site calibration solutions for motorized LiDAR systems. To the best of our knowledge, existing methods still struggle to achieve accurate and efficient on-site calibration of motorized LiDAR systems, highlighting the need for a more effective solution. Addressing these challenges, we propose LiMo-Calib, an on-site fast calibration method that leverages the inherent geometric features in raw LiDAR scans without the need for external targets. To ensure both efficiency and accuracy, we strategically select the most suitable primitives based on their normal distribution, which accelerates optimization while maintaining comparable calibration precision. Additionally, to enhance robustness across diverse scenarios, we introduce a weighted mechanism that evaluates the quality of local plane fitting during the correspondence construction process. This approach prioritizes high-quality features, improving the reliability and adaptability of the calibration method on the quadruped robot. The main contributions of LiMo-Calib are summarized as follows: (1) LiMo-Calib selects planar primitives based on their normal distribution to accelerate optimization while preserving calibration accuracy. (2) LiMo-Calib employs a reweighting mechanism to enhance robustness and improve overall accuracy, resulting in precise and reliable panoramic 3D sensing. (3) LiMo-Calib is evaluated on the in-house LiDAR-motor panoramic 3D sensing system. We further demonstrate the accuracy improvements of the LIO on the panoramic 3D sensing system using the calibrated parameters, which demonstrates efficiency and accuracy in real-world applications. II Related Work Sensor calibration is a fundamental requirement for robust perception in autonomous systems. Early efforts in multi-sensor calibration, such as LiDAR-IMU, LiDAR-camera, and LiDAR-motor calibration, predominantly focused on controlled, offline environments. Many established methods in these domains rely on artificial targets or pre-structured settings to establish precise correspondences (e.g., [ 19 , 20 , 21 ] ). While target-based approaches can achieve high accuracy, their dependence on manually deployed calibration objects limits their practicality for on-site applications, particularly in unstructured environments where such setups are often infeasible. To overcome the limitations of target-based calibration, recent research has focused on targetless methods that leverage naturally occurring features such as edges, planes, and semantic boundaries from raw sensor data [ 13 , 22 , 16 ] . These approaches extract dense geometric or semantic features and employ complex nonlinear optimization to align sensor measurements, but they incur substantial computational overhead and are highly sensitive to scene structure; their performance may degrade in environments with sparse salient features or overly complex geometry. Moreover, many of these methods assume static or slowly varying conditions, limiting their real-time applicability on mobile platforms. For instance, Skaloud and

Read the complete text