An improved self-adaptive Kalman filter for underwater integrated navigation system based on DR

Owing to the atrocious oceanic operating environment of autonomous underwater vehicles(AUV), there are many uncertainties of sensors data with big noises, and high rate of wild points, especially underwater acoustic sensors. An improvement Self-adaptive Kalman filter(SAKF) is designed with forgetting factor introduced, and the optimal forgetting factor is given by the predictive residual error method. In addition, some methods are adopted to avoid probable filter divergence caused by the low estimation accuracy of noise statistic characteristics. The improved SAKF is applied to integrated navigation system of AUV based on dead-reckoning(DR). The integrated navigation system architecture of AUV is described and DR navigation algorithm and motion model are presented in detail. Experimental results show that the improved SAKF is effective, and the navigation system is reliable and feasible.