基于分层马尔可夫决策过程的AUV全局路径规划研究

自主路径规划是自治式水下机器人(AUV)自主能力的重要体现,是保障AUV在大范围复杂海洋环境中自主完成使命作业的关键技术之一。提出了基于马尔可夫决策过程的路径规划方法;并建立了基本的马尔可夫决策模型和结合状态聚类的分层马尔可夫决策模型,同时给出了两种规划的仿真实验及结果分析。实验证明,此类方法能够很好地求解大范围复杂环境内AUV的二维路径规划问题。

Global Path Planning for AUV Based on Hierarchical Markov Decision Processes

The autonomous path planning is an important ability for AUV, which is a crucial factor for ensuring and guiding AUV to accomplish its long-term navigating and autonomous mission in large range and wicked oceanic area. A path planning was preferred based on Markov decision processes (MDP). The basic MDPs model and hierarchical MDPs based on the method of clustering states were introduced. Two simulation experiments and results were given. This kind of method is good at resolving the problem of path planning for AUV through simulation.