Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
80 tokens/sec
GPT-4o
59 tokens/sec
Gemini 2.5 Pro Pro
43 tokens/sec
o3 Pro
7 tokens/sec
GPT-4.1 Pro
50 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Towards High-Performance Solid-State-LiDAR-Inertial Odometry and Mapping (2010.13150v3)

Published 25 Oct 2020 in cs.RO

Abstract: We present a novel tightly-coupled LiDAR-inertial odometry and mapping scheme for both solid-state and mechanical LiDARs. As frontend, a feature-based lightweight LiDAR odometry provides fast motion estimates for adaptive keyframe selection. As backend, a hierarchical keyframe-based sliding window optimization is performed through marginalization for directly fusing IMU and LiDAR measurements. For the Livox Horizon, a newly released solid-state LiDAR, a novel feature extraction method is proposed to handle its irregular scan pattern during preprocessing. LiLi-OM (Livox LiDAR-inertial odometry and mapping) is real-time capable and achieves superior accuracy over state-of-the-art systems for both LiDAR types on public data sets of mechanical LiDARs and in experiments using the Livox Horizon. Source code and recorded experimental data sets are available at https://github.com/KIT-ISAS/lili-om.

Tightly-Coupled LiDAR-Inertial Odometry and Mapping: A Detailed Exploration

The paper "Towards High-Performance Solid-State-LiDAR-Inertial Odometry and Mapping" by Li, Li, and Hanebeck presents an advanced approach combining LiDAR and inertial measurements to improve odometry and mapping accuracy. This paper introduces a novel tightly-coupled scheme designed for both solid-state and mechanical LiDAR systems, notably achieving real-time capabilities with superior performance in various experimental scenarios.

Key Innovations and Methodology

The authors propose a system named LiLi-OM, which implements a tightly-coupled LiDAR-inertial odometry and mapping approach. This system's frontend consists of a feature-based LiDAR odometry that quickly estimates motion for adaptive keyframe selection. In contrast, the backend uses a hierarchical sliding window optimization which marginalizes previous states while directly fusing Inertial Measurement Unit (IMU) and LiDAR data.

The innovation extends particularly to the handling of solid-state LiDARs like the Livox Horizon, which exhibits non-repetitive, irregular scan patterns. This paper introduces a sophisticated feature extraction methodology tailored for this LiDAR type, enabling the system to exploit Livox Horizon’s small Field of View (FoV) efficiently.

Results and Evaluation

The evaluation involves both publicly available datasets and experimental data, demonstrating the system's capabilities against several state-of-the-art benchmarks. LiLi-OM consistently outperforms existing methods, such as LOAM, LIO-SAM, and others, in terms of tracking accuracy and runtime efficiency. This is validated across diverse scenarios, including urban navigation and complex environments involving large loops.

Quantitatively, the presented system achieves best-in-class accuracy, notably reducing errors significantly on datasets like UTBM and UrbanNav. Furthermore, LiLi-OM verifies performance with a scalable and universally applicable backend, maintaining real-time processing capabilities—a substantial advancement in the domain.

Theoretical and Practical Implications

The proposed system's tightly-coupled framework ensures minimal information loss and improved accuracy over traditional decoupled methods. By integrating a robust feature extraction strategy for solid-state LiDARs, the research opens new possibilities in the use of cost-effective LiDAR systems for egomotion estimation and mapping tasks.

Practically, the deployment of a low-cost sensor suite without sacrificing performance shows substantial promise for autonomous applications requiring real-time processing on portable platforms. The development of such a system could significantly advance sectors like mobile robotics, autonomous driving, and unmanned aerial vehicles, offering comprehensive mapping solutions and robust localization capabilities.

Future Prospects

The paper paves the way for further research into enhancing map representation efficiency, particularly in large-scale environments. There is potential for integrating advanced mapping techniques like TSDF for volumetric mapping or geometric primitives for improved memory and computation efficiency. Furthermore, extending the adaptability of the system to aggressive six-DoF egomotion scenarios could broaden its applicability.

Conclusion

This paper contributes significantly to the field of robotics and sensor fusion by offering a highly efficient, cost-effective solution for LiDAR-inertial odometry and mapping. The exploration of solid-state LiDARs within the tightly-coupled framework marks a significant enhancement in odometry accuracy and mapping consistency, providing a reliable fusion strategy applicable to diverse robotic applications.

User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (3)
  1. Kailai Li (11 papers)
  2. Meng Li (244 papers)
  3. Uwe D. Hanebeck (36 papers)
Citations (172)