Research On Docking Oriented AUV Navigation And Path Planning

In recent years, the autonomous underwater vehicle (AUV) plays a more and more important role in the field of oceanic engineering. On one hand, the increasingly complex scenarios demand for more autonomy and longer endurance of AUVs. On the other hand, due to the limitations of the onboard battery and inefficient underwater communication, most AUVs now are only capable of short term operations and still need frequent deployment and recovery. As an effective solution to this contradiction, the docking operation allows a vehicle to recharge its battery, exchange information and accept new missions. Basing on a small AUV developed in our laboratory, this dissertation investigates the docking navigation technique, path planning and system implementation. The major research work is as follows:This paper develops a modified Cubature Kalman Filter (MCKF) for the three dimentional navigation model built on the docking motion of the AUV, which greatly reduces the amount of computation. To solve the problem of delayed measurement in the scenario of USBL-aided AUV docking operation, this paper introduces several delayed measurement filter algorithms then compares them by simulation. And to avoid the impact of outliers on the navigation accuracy, the outlier-rebust MCKF algorithm is proposed, which combines the MCKF with the residual-based outlier detection method and further improves the filtering performance and stability.This paper studies a path planing algorithm based on Dubins path, which takes the motion characteristics of underactuated AUVs into account. Considering there could be dangerous zones in the complex underwater terrain, this paper proposes a novel 2D docking path planning algorithm based on graph theory, which uses the known terrain information and guarantees to generate the optimal Dubins path in polynomial time complexity. Then on this basis an empirical suboptimal extension is presented for the 3D cases.This paper also implements the whole navigation and communication system for the AUV docking scenario, basing on the small AUV platform and an open source software named MOOS, short for Mission Oriented Operating Suite. To overcome the shortcomings of the communication mechanism in the MOOS framework, an improved scheme based on the Push model is developed. Finally, this paper implements an MCKF-based navigator under the MOOS framework and preliminarily verifies its practicability by both simulation and a pool test.