Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

From Underground Mines to Offices: A Versatile and Robust Framework for Range-Inertial SLAM

Lorenzo Montano-Oliván, Julio A. Placed, Luis Montano, María T. Lázaro

TL;DR

LG-SLAM addresses the need for robust, globally consistent SLAM across diverse environments by fusing LiDAR range data, IMU, and GNSS in a two-stage pose-graph optimization. Its key contributions are modular adaptability to different sensors, submap-based local geometry, a robust loop-closure mechanism, and efficient global optimization enabled by GPU acceleration. Empirical results on public datasets and real-world deployments show sub-decimeter to decimeter accuracy (≈12–19 cm ATE) and real-time performance, outperforming several state-of-the-art methods and demonstrating reliability in GNSS-denied and feature-sparse settings. The framework holds promise for wide applicability in autonomous robotics, construction, and surveying where versatile sensing and online mapping are critical.

Abstract

Simultaneous Localization and Mapping (SLAM) is an essential component of autonomous robotic applications and self-driving vehicles, enabling them to understand and operate in their environment. Many SLAM systems have been proposed in the last decade, but they are often complex to adapt to different settings or sensor setups. In this work, we present LiDAR Graph-SLAM (LG-SLAM), a versatile range-inertial SLAM framework that can be adapted to different types of sensors and environments, from underground mines to offices with minimal parameter tuning. Our system integrates range, inertial and GNSS measurements into a graph-based optimization framework. We also use a refined submap management approach and a robust loop closure method that effectively accounts for uncertainty in the identification and validation of putative loop closures, ensuring global consistency and robustness. Enabled by a parallelized architecture and GPU integration, our system achieves pose estimation at LiDAR frame rate, along with online loop closing and graph optimization. We validate our system in diverse environments using public datasets and real-world data, consistently achieving an average error below 20 cm and outperforming other state-of-the-art algorithms.

From Underground Mines to Offices: A Versatile and Robust Framework for Range-Inertial SLAM

TL;DR

LG-SLAM addresses the need for robust, globally consistent SLAM across diverse environments by fusing LiDAR range data, IMU, and GNSS in a two-stage pose-graph optimization. Its key contributions are modular adaptability to different sensors, submap-based local geometry, a robust loop-closure mechanism, and efficient global optimization enabled by GPU acceleration. Empirical results on public datasets and real-world deployments show sub-decimeter to decimeter accuracy (≈12–19 cm ATE) and real-time performance, outperforming several state-of-the-art methods and demonstrating reliability in GNSS-denied and feature-sparse settings. The framework holds promise for wide applicability in autonomous robotics, construction, and surveying where versatile sensing and online mapping are critical.

Abstract

Simultaneous Localization and Mapping (SLAM) is an essential component of autonomous robotic applications and self-driving vehicles, enabling them to understand and operate in their environment. Many SLAM systems have been proposed in the last decade, but they are often complex to adapt to different settings or sensor setups. In this work, we present LiDAR Graph-SLAM (LG-SLAM), a versatile range-inertial SLAM framework that can be adapted to different types of sensors and environments, from underground mines to offices with minimal parameter tuning. Our system integrates range, inertial and GNSS measurements into a graph-based optimization framework. We also use a refined submap management approach and a robust loop closure method that effectively accounts for uncertainty in the identification and validation of putative loop closures, ensuring global consistency and robustness. Enabled by a parallelized architecture and GPU integration, our system achieves pose estimation at LiDAR frame rate, along with online loop closing and graph optimization. We validate our system in diverse environments using public datasets and real-world data, consistently achieving an average error below 20 cm and outperforming other state-of-the-art algorithms.
Paper Structure (12 sections, 3 equations, 8 figures, 3 tables)

This paper contains 12 sections, 3 equations, 8 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methods
  4. System Overview
  5. Feeding Suitable Input Data
  6. Robust Odometry in a Sliding Graph Fashion
  7. Keyframe Selection and Local Consistency
  8. Global Consistency and Optimization
  9. Experiments
  10. Dataset Benchmark
  11. Real-world Deployment
  12. Conclusion

Figures (8)

  • Figure 1: Visualization of the georeferenced 3D reconstruction and pose-graph (blue) built in an urban area ($\sim9$ km). LG-SLAM successfully operates in real time in a large environment, integrating heterogeneous measurements within a robust loop-closure validation process and producing accurate, consistent results even in GNSS-shadowed and featureless regions. The zoomed views in the figure highlight these key capabilities of our system.
  • Figure 2: Overview of LG-SLAM.
  • Figure 3: Comparison of a scan before (blue) and after (red) de-skewing. Distortion increases with the sweep motion the and the point distance: it is maximum in the lower left region.
  • Figure 4: Graph architecture of LG-SLAM. Multiple sensor measurements are integrated into a pose-graph framework that combines pre-integration, GNSS, scan-to-submap and submap-to-submap matching constraints. Submap formation and loop closing are also represented.
  • Figure 5: Pose-graphs (orange) and 3D maps built by LG-SLAM in Underground (left) and Stairs (right) from NCE dataset zhang21.
  • ...and 3 more figures