Adaptive Robust Unscented Kalman Filter for AUV Acoustic Navigation

Area of Science:

• Robotics
• Marine Engineering
• Signal Processing

Background:

• Autonomous underwater vehicle (AUV) acoustic navigation faces challenges from unknown system noise and observation errors in complex marine environments.
• Classical unscented Kalman filter (UKF) algorithms struggle with dynamic model biases and gross errors, limiting navigation accuracy.
• Robustness and adaptability are crucial for reliable AUV positioning in real-world conditions.

Purpose of the Study:

• To develop an adaptive robust UKF for AUV acoustic navigation that addresses limitations of classical UKF.
• To improve system noise estimation and mitigate the impact of gross errors in acoustic observations.
• To enhance the accuracy and stability of AUV navigation systems.

Main Methods:

• Proposed an adaptive robust UKF integrating the Sage-Husa filter for online system noise compensation.
• Implemented a robust estimation technique using Huber's equivalent weight function to control gross error influence.
• Validated the algorithm through simulated long baseline positioning and real marine experimental data.

Main Results:

• The adaptive UKF effectively estimates time-varying system noise, preventing negative definite noise variance matrices.
• The proposed adaptive robust UKF significantly reduces the impact of gross errors while adjusting system noise.
• Experimental results demonstrate improved accuracy and stability in AUV acoustic navigation.

Conclusions:

• The adaptive robust UKF offers a superior solution for AUV acoustic navigation compared to traditional methods.
• The algorithm's ability to handle system noise and gross errors enhances reliability in challenging marine environments.
• This approach contributes to more precise and stable underwater navigation for AUVs.