Combining DVL-INS and Laser-Based Loop Closures in a Batch Estimation Framework for Underwater Positioning (2307.04885v1)

Abstract: Correcting gradual position drift is a challenge in long-term subsea navigation. Though highly accurate, modern inertial navigation system (INS) estimates will drift over time due to the accumulated effects of sensor noise and biases, even with acoustic aiding from a Doppler velocity log (DVL). The raw sensor measurements and estimation algorithms used by the DVL-aided INS are often proprietary, which restricts the fusion of additional sensors that could bound navigation drift over time. In this letter, the raw sensor measurements and their respective covariances are estimated from the DVL-aided INS output using semidefinite programming tools. The estimated measurements are then augmented with laser-based loop-closure measurements in a batch state estimation framework to correct planar position errors. The heading uncertainty from the DVL-aided INS is also considered in the estimation of the updated positions. The pipeline is tested in simulation and on experimental field data. The proposed methodology reduces the long-term navigation drift by more than 30 times compared to the DVL-aided INS estimate.

• The paper presents a batch estimation framework that fuses DVL-INS outputs with laser loop closures to significantly reduce drift errors.
• It employs semidefinite programming to recover raw sensor data and integrates these measurements within an enhanced Kalman filter-based pose graph.
• Experimental results show over 30-fold drift reduction, demonstrating robust performance in real-world AUV navigation tasks.

Underwater Positioning: Integrating DVL-INS and Laser-based Loop Closures

The presented paper investigates an innovative approach to enhancing underwater navigation by integrating Doppler velocity log-aided inertial navigation systems (DVL-INS) and laser-based loop closures within a batch estimation framework. The paper seeks to address a prominent issue in subsea navigation: the gradual drift in position estimation over time due to sensor noise and biases, even with acoustic aiding. This approach is particularly relevant for autonomous underwater vehicles (AUVs) which perform tasks such as subsea metrology and oceanographic surveys where precise navigation is critical.

Methodology and Contributions

The authors propose a methodology to augment the DVL-INS position drift correction built upon estimating raw sensor measurements from DVL-INS outputs using a convex optimization approach, specifically semidefinite programming (SDP). The paper makes several prominent contributions:

• Estimation of Raw Sensor Measurements: The authors introduce a strategy using SDP to estimate raw interoceptive and exteroceptive measurements by considering the DVL-INS outputs as a black-box system. This estimation recovers the missing raw data for inclusion in a batch state estimation framework.
• Integration with Laser-based Loop Closures: The estimated measurements are then fused with laser-derived loop closure measurements. These loop closures are generated by matching features in scanned data, providing relative position corrections between non-sequential time steps.
• Covariance Estimation Incorporating Heading Uncertainty: This paper considers the heading uncertainties inherent to DVL-INS in formulating the process model, augmenting the covariance calculations to provide more robust position updates.

The methodology involves a detailed treatment of the Kalman filter framework and develops an innovative pose graph approach to facilitate the integration of diversely sourced navigation data.

Results and Implications

In terms of performance, the approach reported a significant reduction in navigation drift—by more than 30 times compared to standard DVL-INS operation—over trajectories typically subjected to drift issues. Simulations mirrored real-world conditions, confirming that the methodology proficiently bounds drift errors as corrected by loop-closure measurements. Furthermore, experimental tests with real field data reinforced these results, affirming the algorithm's practical utility.

Theoretical Implications: The proposed paradigm exemplifies how merging data from various sensors can enhance traditional navigation systems like DVL-INS, offering a framework adaptable to other domains requiring precise navigation solutions.

Practical Implications: Deploying the methodology with existing AUV systems can significantly enhance operation accuracy without extensive hardware changes, simply through improved data integration techniques.

Future Directions

The research opens avenues for further work, primarily focusing on enhancing the robustness of covariance estimates, extending methodologies to three-dimensional environments, and fine-tuning the process model in response to dynamic underwater environments. The intricacies of integrating diverse sensor data reflect broader trends in AI where multi-modal data fusion targets improved decision-making capabilities.

In conclusion, this paper offers a significant contribution to underwater navigation by effectively integrating advanced computational techniques to extend the capabilities of existing sensor systems. This approach presents a substantive gain in subsea operation accuracy, marking a noteworthy advancement in inertial navigation systems enhanced by loop-closure measurements.