Application of Modified EKF Based on Intelligent Data Fusion in AUV Navigation

Effective and accurate navigation is critical for Autonomous Underwater Vehicle (AUV). Extended Kalman Filter (EKF) is widely used in AUV navigation system to ensure precise localization. Nevertheless, unknown deviation from low-cost sensors such as Attitude and Heading Reference System (AHRS) and Doppler velocity log (DVL) may have an adverse impact on AUV navigation. This paper proposed an intelligent data fusion method utilizing deep neural networks to modify the EKF to compensate for the deviation. This proposed approach has been tested on the experimental datasets acquired by our own research platform, Sailfish AUV-210, during lake trials at Menlou Reservoir and sea trials at Tuandao Bay. The results indicated that the performance of presented algorithm is significantly superior to EKF and even better than Unscented Kalman Filter (UKF), revealing that the proposed algorithm successfully improving the accuracy of navigation system.

[1]  Richard A. Brown,et al.  Introduction to random signals and applied kalman filtering (3rd ed , 2012 .

[2]  Rudolph van der Merwe,et al.  The square-root unscented Kalman filter for state and parameter-estimation , 2001, 2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.01CH37221).

[3]  Wojciech Zaremba,et al.  Recurrent Neural Network Regularization , 2014, ArXiv.

[4]  Jürgen Schmidhuber,et al.  Deep learning in neural networks: An overview , 2014, Neural Networks.

[5]  Guoqiang Mao,et al.  Design of an Extended Kalman Filter for UAV Localization , 2007, 2007 Information, Decision and Control.

[6]  John J. Leonard,et al.  Pure range-only sub-sea SLAM , 2003, 2003 IEEE International Conference on Robotics and Automation (Cat. No.03CH37422).

[7]  Sajad Saeedi,et al.  AUV Navigation and Localization: A Review , 2014, IEEE Journal of Oceanic Engineering.