AUV Integrated Navigation Algorithm Study And System Implementation

Autonomous Underwater Vehicle (AUV) has been a hot topic in the field of oceanic engineering in recent years; it has increasing applications in underwater environmental surveillance, offshore oil engineering operations, underwater search and mapping, and real-time warzone guarding. When equipped with various sensor devices, AUV can be deployed to collect relevant underwater information effectively.The navigation technique is a key for the development of AUV. Considering the factors of long operation time, covertness, requirement for high level reliability and working under complicated environments, it is a great challenge for oceanic engineers and scientists to design an efficient AUV navigation system. Error of the traditional dead reckoning navigation system based on a single sensor grows quite fast with time; therefore people often refer to integrated navigation algorithms. A commonly-used one is the so-called Extended Kalman Filter (EKF), however, its performance degrades when it comes to a nonlinear situation. In this thesis, we introduce the Unscented Kalman Filter (UKF), traditional Particle Filter (PF) and Rao-Blackwellised Particle Filter (RBPF) to replace the EKF. On the basis of AUV dynamic behavior modeling, we have established an implementation framework and detailed steps of AUV navigation for the above-mentioned algorithms. By analyzing the numeric simulations'and experiments'result, we deduce the advantages and drawbacks of these navigation algorithms.In this thesis, the "Dolphin 1" AUV built by Zhejiang University, particularly its navigation system, is also introduced. The process of developing the real-time EKF-based navigation implementation on DSP is discussed. The navigation results from the TI C6713 DSP firmware has an agreeable performance with that of the computer language, which meets the requirement for precision and real-time AUV operation.