Polar Grid Navigation Algorithm for Unmanned Underwater Vehicles

Zheping Yan, Lu Wang, Wei Zhang, Jiajia Zhou, Man WangSensors 2017 · DOI

TL;DR

This paper proposes a novel polar grid navigation algorithm for Unmanned Underwater Vehicles (UUVs) to overcome the limitations of traditional navigation systems in polar regions. By integrating a Strapdown Inertial Navigation System (SINS) with OCTANS and Doppler Velocity Log (DVL) and employing a modified fuzzy adaptive Kalman filter (MFAKF), the algorithm achieves improved navigation accuracy in challenging polar environments.

What Problem Does This Paper Address?

Traditional navigation systems, particularly those relying on the Earth-centered Earth-fixed (ECEF) frame, struggle in polar regions due to the rapid convergence of Earth meridians. This convergence leads to significant calculation overflows and sharply increasing errors, making accurate heading and position estimation difficult. Furthermore, UUVs operating in polar regions face unique challenges, including the unavailability of GPS signals underwater, complex environmental conditions, and the need for high autonomy without human intervention. Existing navigation algorithms designed for non-polar regions are not directly applicable to UUVs in these demanding polar environments.

Methods

The core of the proposed solution is a polar grid navigation algorithm. This algorithm utilizes a grid frame as the navigation frame, which is designed to mitigate the issues caused by meridian convergence. The navigation system is an integrated approach combining:

  • Strapdown Inertial Navigation System (SINS): Provides high autonomy and stability, forming the primary navigation source.
  • OCTANS: A gyrocompass and motion sensor used to measure the UUV's attitude.
  • Doppler Velocity Log (DVL): Measures the UUV's velocity relative to the seafloor.

To fuse the data from these sensors and improve navigation accuracy, a modified fuzzy adaptive Kalman filter (MFAKF) is employed. This filter is an enhancement of the traditional Adaptive Kalman Filter (AKF). Key features of the MFAKF include:

  • Modified Fuzzy Logic: A fuzzy inference system (FIS) is used to adaptively adjust the measurement noise covariance (Rk). This adaptation helps the filter respond better to changing underwater environments.
  • T-S Fuzzy Logic: Utilized in the adaptive regulation of noise covariance, aiming to keep residual errors close to zero.
  • Neglecting Negative Terms: Simplifies the filter while maintaining the positive definiteness of the covariance matrices.

The paper also details the derivation of error models for the SINS, OCTANS, and DVL within the context of the polar grid frame.

Datasets

Not discussed in the paper. The paper relies on simulations and experimental data collected from a UUV built by the authors.

Key Insights and Ideas

  • Grid Frame Superiority in Polar Regions: The grid frame effectively circumvents the meridian convergence problem that plagues traditional navigation frames in high-latitude areas, leading to more stable and accurate navigation.
  • MFAKF Outperforms AKF: The fuzzy adaptive nature of the MFAKF allows it to better handle environmental variations and sensor noise, resulting in significantly improved estimation accuracy compared to a standard AKF.
  • Integrated Navigation is Crucial: Combining SINS with OCTANS and DVL provides complementary information, enhancing the overall robustness and accuracy of the navigation system, especially in environments where single sensors might fail or be unreliable.
  • Model 1's Effectiveness: The proposed polar grid navigation algorithm (Model 1) demonstrated superior performance over traditional SINS models (Model 2 and Model 3) in both simulations and experiments within the polar region.

Key Results

  • Simulation Results: The proposed polar grid navigation algorithm (Model 1) showed more accurate performance than Model 2 and Model 3. The MFAKF demonstrated better performance than the AKF, with RMS estimation errors for attitude, velocity, and position being lower. Specifically, RMS errors for MFAKF were less than 9.0768°, 0.0256 m/s, and 556.3639 m, respectively, compared to AKF's 14.3786°, 0.0912 m/s, and 573.4550 m.
  • Experimental Results: Similar to simulations, experimental results showed that Model 1 was superior to Model 2 and Model 3. The MFAKF also exhibited better navigation accuracy than the AKF. RMS errors for MFAKF were less than 11.0893°, 0.0376 m/s, and 551.6560 m, compared to AKF's 15.1294°, 0.0909 m/s, and 556.5706 m.

Limitations and Future Work

  • Limitations:
    • The paper acknowledges that the transversal SINS algorithm (Model 3) has principle errors as it ignores the Earth's ellipse.
  • While the MFAKF improves accuracy, the paper doesn't explicitly detail the computational overhead introduced by the fuzzy logic.
  • Experimental validation was conducted in a rectangular pool in a non-polar region for semi-physical simulation, with polar region data being a combination of practical and simulated data. Full-scale polar experiments are not detailed.
  • Future Work: The paper suggests that the proposed polar grid navigation algorithm for UUVs can be effectively used in the polar region. The MFAKF's ability to adjust to changing environments implies its potential for broader applications in dynamic underwater scenarios. Further research could involve full-scale polar expeditions to validate the algorithm under real-world conditions and explore its performance in more complex underwater terrains and dynamic currents.

Summary updated

BibTeX
@article{yan2017polar,
  author = {Zheping Yan and Lu Wang and Wei Zhang and Jiajia Zhou and Man Wang},
  title = {{Polar Grid Navigation Algorithm for Unmanned Underwater Vehicles}},
  journal = {Sensors},
  year = {2017},
  doi = {10.3390/s17071599},
}
Figure