SIA OpenIR  > 海洋信息技术装备中心
对抗环境下多水下机器人协同围捕方法研究
Alternative TitleResearch on the cooperative pursuit method of multiple underwater vehicles in an adversarial environment
贾庆勇
Department海洋信息技术装备中心
Thesis Advisor封锡盛 ; 徐红丽
Keyword多水下机器人 运动目标 协同搜索 模型预测控制 协同围捕
Pages155页
Degree Discipline机械电子工程
Degree Name博士
2019-11-28
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract多水下机器人(Autonomous Underwater Vehicle,AUV)协同围捕是研究多机器人之间协调与对抗的重要方向,也是人工智能领域的研究热点之一。如何有效地控制多个水下机器人在隐蔽的水下环境中协同执行围捕任务在民事和军事领域具有重要的研究价值和现实意义。本文重点针对协同围捕过程中动态目标搜索、静态目标围捕以及动态目标围捕方法进行了研究。本文的主要研究内容如下:(1)针对水下大范围环境中单个运动目标的搜索问题,提出了一种多水下机器人分布式协同搜索策略,用于减少统计意义下搜索目标的平均时间,提高发现目标的概率。从基本搜索问题出发,深入分析了多水下机器人协同搜索问题的过程和要素,建立了多水下机器人系统平台、传感器、目标和网络通信的分布式协同搜索问题框架。通过关键问题分解,将协同搜索划分为环境信息建模和协同任务决策两个关键子问题。深入研究了基于搜索图的环境模型,重点研究了搜索图的探测更新、通信更新和预测更新问题。基于模型预测控制(MPC)的思想,研究了多水下机器人分布式模型预测控制的建模方法,将集中式优化决策问题转化为多个水下机器人子系统的分布式优化决策问题,减小了多水下机器人的求解规模,并基于粒子群优化(PSO)算法实现了多水下机器人分布式优化决策问题的求解。最后,通过蒙特卡洛仿真实验验证了所提协同搜索策略的有效性。(2)针对目标的运动状态信息以及对抗策略,提出了一种基于快速随机搜索树(RRT)的多水下机器人协同围捕策略。该策略将追逃博弈划分为两个阶段。在第一阶段,目标由运动状态变为静止状态,多水下机器人基于RRT的方法对静态目标实现隐蔽性包围,从而避开目标的探测范围,到达期望的预包围位置。在第二阶段,当目标感知到多水下机器人进入目标探测范围时,目标则开始逃跑;提出一种多水下机器人协同围捕控制律对运动目标进行跟踪和围捕。仿真结果表明,多水下机器人利用该协同围捕策略可以有效实现静态目标的隐蔽包围,并对试图逃跑的目标实现稳定跟踪。(3)针对水下单个运动目标的围捕问题,提出了一种基于队形控制的多水下机器人协同围捕策略。首先分析了多水下机器人协同围捕过程和要素,建立了多水下机器人协同围捕问题框架。通过关键问题分解,将协同围捕问题划分为协同围捕问题建模、围捕队形设计和协同控制器设计三个关键子问题。然后重点基于有限状态机理论建立了多水下机器人协同围捕模型,基于不同的环境信息设计了不同的协同围捕队形,并分析了每一种围捕队形的意义和价值。针对不同的围捕队形,基于比例控制的思想设计了多水下机器人协同控制协议。仿真和湖试试验结果表明了该协同围捕策略的可行性和有效性。本文的主要创新点包含以下几个方面:(1)本文提出一种新颖的搜索思路,用于多水下机器人在大范围海域中协同搜索运动目标。该搜索思路创新性的提出了一种基于目标初始信息的搜索图更新机制,提出了一种目标运动范围的预测方法,建立了均匀概率分布模型,并给出了搜索图的目标概率更新公式。同时,将预测控制思想用于多水下机器人优化决策中。该搜索思路有效降低了多水下机器人搜索运动目标的盲目性,提高了搜索运动目标的概率,减少了发现目标的时间,为解决多水下机器人协同围捕问题提供了一定的理论支撑。(2)本文针对水下单个静态目标围捕问题,提出了一种新颖的多水下机器人协同围捕策略,该协同围捕策略解决了水下环境中包围静态目标的隐蔽性问题,将多水下机器人协同围捕静态目标问题,分解为单个水下机器人的路径实时重规划问题,降低了问题的复杂度,同时解决了运动过程中的避障问题。(3)本文针对单个运动目标围捕问题,提出了一种新的多水下机器人协同围捕策略,并将有限状态机理论应用于多水下机器人围捕过程中。同时,针对围捕过程中的各围捕状态设计了相应的围捕策略,创新性的提出了半圆形包围策略和圆形包围策略,实现了多水下机器人对运动目标隐蔽包围。该协同围捕策略对于解决多水下机器人离散动态系统问题具有重要的实际指导意义。
Other AbstractCooperative pursuit of multiple underwater robots is an important means to study the coordination and confrontation between multiple robots, and also one of the research hotspots in the field of artificial intelligence. How to effectively control multiple underwater robots to perform the pursuit task in the hidden underwater environment has important research value and practical significance in civil and military fields. This paper focuses on the methods such as the dynamic target search, the static target pursuit and the dynamic target pursuit in the cooperative pursuit process. Main research contents and innovation points are as follows: (1) Aiming at the search problem of a single moving target in large underwater environment, a distributed cooperative search strategy of multiple underwater robots is proposed, which can reduce the average time of searching target and improve the probability of finding target in statistical sense. Based on the basic search problem, this paper deeply analyzes the process and the elements of cooperative search problem for multiple underwater robots, and establishes the distributed cooperative search problem framework including the system platform model, the sensors model, the target model and the network communication model. The cooperative search problem is divided into the two key sub-problems which are the environment information modeling problem and the cooperative task decision-making problem through breaking down the key problems. The environment model based on search map is deeply studied, and the detection update, the communication update and the prediction update of search map are emphatically studied. Based on the idea of model predictive control (MPC), the modeling approach of distributed model predictive control for multiple underwater robots is studied. The centralized optimal decision-making problem is transformed into the distributed optimal decision-making problem of multiple sub-systems for underwater robots, which reduces the solving scale of multiple underwater robots. The distributed optimal decision problem of multiple underwater robots is solved based on particle swarm optimization (PSO) algorithm. At last, the effectiveness of the proposed cooperative search strategy is verified by the Monte Carlo simulation experiment. (2) Aiming at the motion state information and countermeasures of the target, a cooperative pursuit strategy based on rapidly exploring random tree (RRT) for multiple underwater robots is presented. This strategy divides the pursuit-evasion game into the two stages. In the first stage, the target changes from a moving state to a stationary state, and the static target is encircled secretly by multiple underwater robots based on the method of RRT so as to avoid the detection range of the target and reach the desired pre-encirclement position. In the second stage, when the target senses that some underwater robot entered into the detection range of the target, the target start to escape. A cooperative pursuit control law of multiple underwater robots is proposed for tracking and pursuing the moving target. The simulation results shows that the proposed cooperative pursuit strategy can effectively implement the concealment encirlement of the static target, and implement stable tracking of the target trying to escape. (3) Aiming at the pursuit problem of a single moving target under the water, a cooperative pursuit strategy based on formation control is proposed for multiple underwater robots. Firstly, the process and elements of cooperative pursuit for multiple underwater robots is analyzed and the cooperative pursuit framework is also established. Through the decomposition of the key problems, the collaborative encirclement problem is divided into three key sub-problems: the collaborative encirclement problem modeling, the pursuit formation design and the controller design. Then, based on the theory of finite state machine, the model of cooperative pursuit for underwater vehicles is established, and different cooperative encirclement formations are designed based on different environmental information, and the significance and value of each pursuit formation are analyzed. Based on the idea of proportional control, the cooperative control protocol of multiple underwater robots is designed for different pursuit formation. The simulation and lake test results show the feasibility and effectiveness of the proposed strategy. The main innovation points of this paper include the following aspects: (1) This paper presents a novel search method for multiple underwater robots to search for a moving target in a large sea area. This search idea innovatively proposes a search map updating mechanism based on the initial information of the target, proposes a prediction method of the target motion range, establishes the uniform probability distribution model, and gives the target probability updating formula of the search map. At the same time, the predictive control method is applied to the optimal decision-making of multiple underwater robots. This search idea effectively reduces the blindness of the underwater robots in searching the moving target, improves the probability of searching the moving target, reduces the time to find the target, and provides some theoretical support for solving the problem of cooperative pursuit for underwater robots. (2) For the pursuit problem of a single static target under the water, and a novel cooperative pursuit strategy for underwater robots is put forward. The cooperative pursuit strategy solves the concealment problem of encircling the static target problem under the water. The pursuit problem of the static target for underwater robots is broken down into the real-time path planning problem of individual underwater robot. The complexity of the problem is reduced, and the obstacle avoidance problem is also solved in the process of movement at the same time. (3) For the pursuit problem of a single moving target under the water, a new cooperative pursuit strategy for underwater robots is presented, and the finite state machine theory is applied to the pursuit process. At the same time, the corresponding pursuit strategies are designed for each pursuit state in the pursuit process, and the semicircular encirclement strategy and circular encirclement strategy are innovatively put forward to realize the covert encirclement of a moving target by underwater robots. This strategy has important practical significance for solving the discrete dynamic system problem of underwater vehicles.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/25938
Collection海洋信息技术装备中心
Recommended Citation
GB/T 7714
贾庆勇. 对抗环境下多水下机器人协同围捕方法研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2019.
Files in This Item:
File Name/Size DocType Version Access License
对抗环境下多水下机器人协同围捕方法研究.(5570KB)学位论文 开放获取CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[贾庆勇]'s Articles
Baidu academic
Similar articles in Baidu academic
[贾庆勇]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[贾庆勇]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.