Improving Self-Consistency in Underwater Mapping Through Laser-Based Loop Closure (Extended) (2301.02297v2)

Abstract: Accurate, self-consistent bathymetric maps are needed to monitor changes in subsea environments and infrastructure. These maps are increasingly collected by underwater vehicles, and mapping requires an accurate vehicle navigation solution. Commercial off-the-shelf (COTS) navigation solutions for underwater vehicles often rely on external acoustic sensors for localization, however survey-grade acoustic sensors are expensive to deploy and limit the range of the vehicle. Techniques from the field of simultaneous localization and mapping, particularly loop closures, can improve the quality of the navigation solution over dead-reckoning, but are difficult to integrate into COTS navigation systems. This work presents a method to improve the self-consistency of bathymetric maps by smoothly integrating loop-closure measurements into the state estimate produced by a commercial subsea navigation system. Integration is done using a white-noise-on-acceleration motion prior, without access to raw sensor measurements or proprietary models. Improvements in map self-consistency are shown for both simulated and experimental datasets, including a 3D scan of an underwater shipwreck in Wiarton, Ontario, Canada.

• The paper demonstrates a robust laser-based loop closure method that extracts precise loop closures from underwater laser scans.
• The paper integrates loop-closure measurements into COTS-based navigation systems without requiring access to raw sensor data.
• The paper shows significant improvements in trajectory accuracy and map self-consistency, validated on both simulated and real-world datasets.

Overview of "Improving Self-Consistency in Underwater Mapping Through Laser-Based Loop Closure"

The paper discusses a methodology aimed at enhancing the self-consistency of underwater maps through a laser-based loop closure approach. The work addresses challenges in underwater vehicle navigation, where precise and self-consistent bathymetric maps are critical for assessing subsea environments and infrastructure. Standard navigation solutions for Autonomous Underwater Vehicles (AUVs) typically involve Commercial Off-the-Shelf (COTS) Inertial Navigation System (INS) with acoustic Doppler velocity log (DVL) aiding. While high-quality systems can achieve extremely low drift rates, they often require expensive and limiting external acoustic localization systems.

The authors propose a novel method for integrating loop-closure measurements into a navigation state estimated by a COTS subsea navigation system, specifically targeting applications lacking access to raw sensor data and proprietary models. The proposed integration employs a white-noise-on-acceleration motion prior, resulting in improved map self-consistency as demonstrated on both simulated and experimental datasets.

Technical Insights and Contributions

1. Loop Closure Utilization: The paper introduces a robust front-end for extracting loop-closure measurements from laser scan data, using advanced point cloud alignment techniques. This approach enables capturing loop closures in complex underwater environments without requiring raw sensor outputs.
2. Integration without Raw Data: A distinctive feature of this method is its capability to seamlessly integrate loop-closure measurements into existing navigation solutions derived from COTS systems. This is achieved without depending on raw measurements, using a batch state estimation framework and factor graph optimization. Notably, the methodology elegantly handles this integration by employing a white-noise-on-acceleration prior to maintain smooth trajectory estimates.
3. Performance Evaluation: The paper evaluates improvements in trajectory accuracy and map self-consistency against both simulated and real-world datasets. Notable improvements are observed, where loop-closure corrections propagate effectively across the trajectory, bounding navigation drift and achieving tighter map alignment.
4. Practical Implications: The methodology enables more accurate subsea mapping and inspection, particularly beneficial in situations where deploying conventional acoustic positioning systems is impractical. By reducing computation without sacrificing accuracy, the approach also holds potential for real-time applications in AUV inspections.
5. Potential for Expansion: Although this paper primarily focuses on laser-based loop closures, the extensible nature of the framework could allow integration of additional sensor modalities, further improving underwater SLAM performance.

Implications and Future Work

The paper sets a significant precedent for enhancing underwater navigation and mapping, with implications spanning marine research, subsea infrastructure monitoring, and autonomous exploration missions. By potentially reducing operational costs and increasing the flexibility of AUV deployments, this research advances the practical application of underwater robotics in diverse scenarios.

Future research could explore automated hyperparameter tuning to optimize performance across various missions, and expand applicability through multi-sensor fusion, including sonar and optical sensors. Further real-world experiments across different marine terrains would clarify the robustness of the proposed methodology under varying environmental conditions.

In conclusion, this work contributes substantially to the field of underwater robotics, providing a scalable solution for achieving precise and contextually aware underwater navigation without dependency on cumbersome, costly external infrastructure.