RLI-SLAM: Fast Robust Ranging-LiDAR-Inertial Tightly-Coupled Localization and Mapping.

Area of Science:

• Robotics
• Computer Vision
• Sensor Fusion

Background:

• Simultaneous Localization and Mapping (SLAM) is crucial for autonomous robot navigation in unknown environments.
• Existing SLAM systems face challenges with sensor drift and real-time accuracy, especially during fast motion.

Purpose of the Study:

• To develop a high-accuracy, fast, and robust SLAM framework for long-term, fast-motion scenarios.
• To improve state estimation and mapping precision in confined and challenging environments.

Main Methods:

• A novel tightly-coupled ranging-LiDAR-inertial SLAM framework (RLI-SLAM) was proposed.
• Ultra-wideband (UWB) ranging and inertial sensors were fused to correct inertial sensor bias and drift.
• An efficient loop closure detection module using an incremental smoothing factor graph was integrated.

Main Results:

• RLI-SLAM demonstrated superior accuracy in state estimation and mapping compared to state-of-the-art systems.
• The framework achieved robust performance even with single ranging measurements and in challenging environments.
• Low computational complexity was maintained, enabling real-time application.

Conclusions:

• The proposed RLI-SLAM framework effectively compensates for LiDAR point cloud distortion during fast motion.
• Tightly fusing UWB ranging and inertial data enhances long-term SLAM accuracy and robustness.
• The system provides high-precision mapping capabilities for autonomous robots in complex, unmapped areas.