面向海洋数据收集和移动目标搜索的多AUV航迹规划

项目名称： 面向海洋数据收集和移动目标搜索的多AUV航迹规划

项目编号： No.61472325

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 其他

项目作者： 崔荣鑫

作者单位： 西北工业大学

项目金额： 83万元

中文摘要： 本项目开展面向海洋数据收集和移动目标搜索的多自主水下航行器（AUV）协同航迹规划研究，主要研究内容和创新点包括：①针对多AUV海洋数据收集问题，提出基于信息一致性的多AUV信息融合方法，并在此基础上设计信息熵驱动的并行快速扩展随机树(RRT)协同航迹规划算法；②考虑目标分布不确定性、传感器探测信息不确定性等因素，在部分可观测马尔科夫决策过程（POMDP）框架下建立多AUV任务规划模型，离线实现面向移动目标搜索的多AUV资源分配和任务规划；③综合考虑目标搜索增益和数据收集增益，将面向移动目标搜索的多AUV航迹规化问题描述为多目标优化问题，提出在线滚动优化的并行RRT算法，实时规划AUV航迹。通过本项目的研究，探索各种不确定性以及约束条件对多AUV系统的影响规律，提高多AUV在数据收集和目标搜索中的自主性。

中文关键词： 自主水下航行器；航迹规划；多目标优化；部分可观测马尔科夫决策过程

英文摘要： In this project, we investigate the cooperative trajectory planning of multiple autonomous underwater vehicles (AUV) for ocean data collection and mobile target searching. The main research focuses and contributions are listed as follows. (1) Consensus based filter will be introduced for the multi-AUV sensor fusion. On the basis of this, entropy driven parallel Rapidly-exploring Random Tree (RRT) trajectory planning method will be proposed for the multiple AUVs. (2) With the target motion uncertainties, the sensor uncertainties and other various uncertainties taken into account, the mathematical model of multi-AUV task allocation will be investigated in the Partially Observable Markov Decision Process (POMDP) framework, which computes the offline resource allocation and task assignment for the multi-AUV system. (3) For the online trajectory planning problem of the multiple AUVs, an online receding optimization based parallel RRT will be presented. Rigorous mathematical proof will be provided for the proposed algorithms. Through the research of this project, we will investigate the effects of the diversified uncertainties and constraints on the performance of multi-AUV system, enhancing the autonomy of the multiple AUVs in the data collection and target searching problem.

英文关键词： Autonomous underwater vehicles;Trajectory planning;Muli-objective optimization;POMDP