The Research Of Adaptive Robust Filtering For Multi-AUV Cooperative Navigation

As a new means of human exploration of the ocean, Autonomous Underwater Vehicles ,not only reliable and safe, reusable, etc., diverse functions can also be achieved by configuring different devices. Single AUV due to its limited volume,can only carry a limited exploration equipment,for a multi AUV of cluster-type. In a multi AUV cooperative formation, the navigation system is the key to the successful implementation task of formation as an important part of the navigation. Cooperative navigation and localization as a new multi AUV navigation formations, not only has a high positioning accuracy can also significantly reduce costs. Due to the special nature of the underwater environment, how to improve robustness and adaptability of the cooperative navigation and localization is essential.This paper presents a collaborative filtering algorithm robust navigation robust navigation based on improved Huber collaborative filtering algorithm. The algorithm can effectively suppress the measurement noises, which in the circumstance of similar Gauss distribution and outliers. Against the question of cooperative navigation and positioning accuracy drops, leading by time varying of measurement noise covariance matrix, proposed to deal with the way of interacting multiple model algorithm.Firstly, introducing the collaborative navigation basic principle and mathematical model for the leader of the case and to establish filtering algorithm under the model (filtering algorithm); describes the use of a single leader at the relative distance and direction to achieve positioning and the use of the motion vector, the two collaborative navigation method, and given the use of the motion vector navigation collaborative filtering algorithm; given the system model and filtering algorithms after considering the currents affect.Secondly, it studies the collaborative navigation robust filter under the measurement noise is similar Gauss distribution. Introducing a Huber estimation method based on the maximum likelihood estimation based, and its application with the collaborative navigation filter algorithm to achieve a robust filtering in the measurement noise is similar Gauss distribution.And then, for the case of multi-AUV Cooperative navigation measurement anomalies,we introduce two methods for outlier identification and analysis through the merits of examples, combined method of identifying outliers gives an improved Huber estimate. In the modified Huber estimate deduced a new collaborative filtering algorithm robust navigation.We discuss the filtering algorithm performance under the circumstance of measurement anomalies.Finally, against the changing collaborative formation work environment, and the time varying of measurement noise covariance matrix measurement characteristics, etc, proposed based on collaborative filtering algorithm Navigation interactive multiple model. Analysis the performance of this algorithm when the measurement noise time varying. The robust filtering algorithm with IMM combination presents a new collaborative navigation robust adaptive filtering algorithm to analyze the effect of the merits of its position.