【作者】 阚维；

【导师】 何波；

【作者基本信息】 中国海洋大学 ， 通信与信息系统， 2011， 硕士

【摘要】 自主式水下机器人(AUV, Autonomous Underwater Vehicle)的上层决策系统是AUV研究的关键技术,在一定程度上决定了AUV的智能化水平。本文旨在探讨一种可靠高效的AUV上层决策系统,使AUV在复杂的水底环境下实现自主导航。上层决策系统的核心是路径规划与决策控制：路径规划是在一定的环境模型中规划出一条从起始状态到目标状态的最优或近似最优无碰路径,决策控制依据规划好的路径对AUV做出运动决策,并且能够处理环境模型中的不确定因素和路径跟踪中出现的误差。本文着重研究了上层决策系统设计中的这两个关键问题,主要研究工作如下：首先对全局路径规划进行了研究,采用了一种基于快速步进算法(FFM, Fast Marching Method)的全局路径规划方法,针对AUV在大范围静态环境完全未知或部分已知的情况下,应用动态划分全局工作环境,以多次局部优化代替一次全局优化的路径规划方法。其次,探讨了一种基于行为的反应式结构决策控制系统,把AUV基本行为封装成理性行为和感性行两大模块。在决策控制系统的避障行为模块中,引入改进人工势场法进行二次局部路径规划,以使AUV在动态环境下能更好地进行实时避障,弥补了单纯路径规划不能应付动态环境的缺点。最后,设计了AUV上层决策系统,此决策系统集成了基于FM算法的路径规划方法与基于行为的反应式决策控制系统的优点,并且采用模块化思想,把决策系统分成任务管理模块、路径规划模块、决策控制模块、定位与地图构建模块(SLAM)等模块,其特点是易于实现与扩展。同时,本文先采用MATLAB语言对决策系统算法进行仿真验证,然后在Microsoft公司的VC++6.0开发环境下,采用C/C++语言实现决策系统。软件运行结果表明,设计的AUV上层决策系统是有效的,能够在复杂的环境下实现AUV的自主导航功能。更多还原

【Abstract】 The upper Decision-making system is the crucial technology for research of modern Autonomous Underwater Vehicle (AUV) and determines AUV’s intelligence level in some extent. This paper aims to explore a reliable and efficient AUV upper Decision-making system which allows AUV to achieve autonomous navigation in complex underwater environment.Path planning and Decision-making control are crucial issues for the AUV’s upper Decision-making system. Path planning is used to extract cost or approximately cost optimal paths form starting position to the end in complex and continuous environments. Decision-making control makes decisions for AUV’s movement based on planned paths, and is capable of handling the uncertainty in environments and correction of path tracking errors. This paper focuses on these two key issues for the design of the upper Decision-making system. The main research work is as follows:Firstly, this paper introduces the research of path planning. Global path planning based on Fast Marching (FM) algorithm is used to extract cost optimal paths from complex underwater environments. Then according to a large scale static unknown or partially unknown environment, this paper proposes a new path planning method which dynamically divides the global environment and uses multiple local optimizations instead of one-time global optimization. And meanwhile, we design a Decision-making control system with reactive structure based on behavior. This system divides AUV’s behavior into two classes-Rational Behavior set and Perceptual Behavior set. Furthermore, in order to avoid dynamic obstacles, in avoidance behavior module of Decision-making control system improved artificial potential field (APF) is used to implement secondary path planning in local environments, which can make up for the disadvantage that the simple path planning couldn’t cope with dynamic environments. Finally, this paper designs a AUV’s upper Decision-making system which integrates the advantage of path planning method based on FM and Decision-making control system with reactive structure based on behavior, and adopts modular method dividing Decision-making system into task management module, path planning module. Decision-making control module, localization and mapping module and so on. which is characterized by easy to implement and expand.And meanwhile, this paper first verifies the effectiveness of AUV’s Decision-making system by MATLAB simulation, then uses C/C++ language to implement this system on Microsoft’s VC++6.0 development platform. Programming running results show that the design of the AUV’s upper Decision-making system is effective, and succeed in implementing AUV’s autonomous navigation with complex environments.更多还原