面向实时避碰的无人水面机器人在线路径规划方法

CORC > 沈阳自动化研究所 > 中国科学院沈阳自动化研究所 > 海洋信息技术装备中心

题名	面向实时避碰的无人水面机器人在线路径规划方法
作者	冷静
学位类别	硕士
答辩日期	2014-05-28
授予单位	中国科学院沈阳自动化研究所
导师	刘健
关键词	无人水面机器人在线路径规划速度避障法混合整数线性规划
其他题名	Online Path Planning for Unmanned Surface Vehicles for real-time obstacle avoidance Online Path Planning for Unmanned Surface Vehicles for real-time obstacle avoidance
学位专业	控制工程
中文摘要	无人水面机器人（USV, Underwater Surface Vehicle）指由携载平台施放，以遥控或自主方式在水面航行的小型无人化、智能化航行器，也称无人水面艇或半潜式无人航行器，近年来成为国外机器人研究领域的一个热点。现阶段，无人水面机器人多采用遥控操作方式，然而智能化的全自主控制是USV未来的发展趋势。全自主控制要求无人水面机器人具有在线应对各种突发状况的能力，特别是在港口、航道等船只密集海域，USV能够自主地规避障碍危险。障碍危险包括水面船只、灯塔、暗堤等，从机器人学角度来看，这些都属于典型的动态环境。针对动态环境，研究在线路径规划方法对于提升无人水面机器人的智能和自主能力具有重要意义，是本文重要的研究内容。本文主要研究内容如下：1.首先对国外USV的避碰体系结构进行分析，它是USV进行在线路径规划的理论框架。在分析无人水面机器人自主避碰特点的基础上，提出基于慎思/反应式体系结构的自主避碰决策系统。自主避碰决策系统分为四个层次，本文重点研究的是在线规划层。它分为三个模块，分别是碰撞危险度评价模块、会遇局面划分及智能决策模块。碰撞危险度评价模块是基于海上船舶避碰中重要的参考量DCPA(最短会遇距离)和TCPA(最小会遇时间)计算出来的碰撞危险度，它是在线路径规划的主要依据。会遇局面划分模块对海洋会遇局面做出划分，以便于做出更安全的决策，这个局面划分必须遵守《1972年国际海上避碰规则公约》。在智能决策模块中，它根据会遇局面与路径规划方法进行优化行为的选择。其中，在线路径规划方法是决策的关键和核心。2.为实现对未知动态障碍物的实时规避，本文分别从障碍区域预测和在线路径规划两个层次开展研究。障碍区域预测是动态障碍物避碰的首要任务，通过对驶来船舶或其他USV的速度和航向的测定，在相对速度空间中，划分速度障碍区和安全区。其中安全区又分为遵守国际海上避碰规则公约(COLREGS)的区域和不符合COLREGS的区域。本文根据速度避障法的原理采用两种改进的速度避障法进行障碍物区域预测，并进行了实验，实验表明第一种方法不能很好遵守COLREGS,第二种能较好遵守COLREGS。速度避障法只能用于局部避开障碍物，它仅能选择避障的速度矢量，但容易陷入局部极小值。选择混合线性规划(MILP)和多目标规划作为全局优化方法。这两种方法都可以将环境的不确定性和USV的动力学作为约束条件，这样就很好考虑了环境的动态性。混合线性规划(MILP)可以根据使目标最优而选择转向，其中有可能为使距离最优而不遵守COLREGS。然而对于有些特定情况不遵守COLREGS也是合理的。多目标规划选择的目标函数可以根据任务的要求选择距离优化函数、速度优化函数和遵守COLREGS目标函数等多个优化目标函数。将速度避障法作为约束条件融入到全局规划器中，使其能够规划既能避开障碍物同时能寻找到目标点的最优解。第一种速度避障法中规划变量为每个规划时间相对目标的速度矢量分量的变化量。第二种速度避障法中规划变量为每个规划时间USV速度大小及角度的变化量。本文对不同优化目标函数进行了实验，实验表明不同目标函数对路径优化有很大影响。3.为了对上述方法进行仿真验证，本文采用Ribcraft USV数学模型进行验证。基于数学模型开发出航行控制器来进行航向和航速的控制，通过控制螺旋桨的转速来控制航速，控制舵角来控制航向，并做了航向和航速的控制实验。最后利用设计的航行控制器在Matlab上进行了在线路径规划方法的仿真实验。分别对多种典型的会遇局面进行了实验仿真。仿真结果验证了该算法的合理有效性。
索取号	TP242.3/L26/2014
英文摘要	With robotic unmanned surface (USV, Underwater Surface Vehicle) research project more diverse in recent years ，it has been a hot field of robotics research abroad. Foreign research and application of positive USV intelligent direction, and our research and application of USV still in its infancy in many key aspects of USV intelligent technology is still very weak. Online path planning for USV in a dynamic environment is a manifestation of an important aspect of USV intelligent and autonomous capabilities. This paper uses the Velocity Obstacle (VO) with the rules of the International Convention for Preventing Collisions at Sea as the avoidance method and hybrid method combining linear programming to achieve the online route planning, and ultimately build the entire collision avoidance decisions. Contents of this paper are as follows: 1. This paper first introduce and analysis foreign USV collision avoidance architecture. It is the theoretical framework for USV online path planning. Then the foreign research status of USV online path planning is summarized. And on this basis, I proposed an avoidance and decision system based on avoiding collision deliberative / reactive hybrid architecture. Collision Avoidance System Planning layer is online planning layer, which is divided into three modules, namely, the collision risk assessment module will encounter the situation and determine the division of modules and intelligent decision-making module. Collision risk assessment module is based on the important maritime ship collision avoidance reference amount DCPA (minimum passing distance) and TCPA (minimum encounter time) is calculated, which is part of the process of collision avoidance and must be considered. The Encounter situation of the marine division of the module is to make division which situation the USV and moving vessel will encounter, in order to make safer decisions, division must comply with this situation and rule of "1972 International Regulations for Preventing Collisions at Sea Convention." In this collision avoidance decision-making system, the online path planning method is the key to make decisions. 2. This paper have researched and analyzed the methods of online path planning for USV. Because USV in the marine environment is complex and dynamic multi unknown, so the VO avoidance method as local obstacle avoidance, it's real good for the environment uncertainty environment, very suitable for high-speed movement of USV obstacle avoidance. Here by setting the steering Preventing Collisions at Sea Convention of the International Rules (COLREGS) very good speed obstacle avoidance into law, making the planned path must comply with COLREGS. VO can only be used locally to avoid obstacles, select a mixed linear programming (MILP) as a global optimization method, which can be uncertain environment dynamics and USV as a constraint, so good considering the environment dynamic. The objective function can be selected from the optimization functions and speed optimization functions according to mission requirements, planning variables change relative to the target amount of planning time each speed. The speed obstacle avoidance method as constraints into MILP global planner which enable them to avoid obstacles while planning both able to find the optimal solution to the target point. 3. In order to carry out the above method simulation, this paper has used the Ribcraft USV mathematical model for validation. Based on the mathematical model developed sailing course and speed controller to control , by controlling the speed of the propeller to control the speed of the USV, steering angle control to control the course of the USV and do the course and speed of the control experiments . Finally, the design of the navigation controller in the Matlab simulation experiments is carried out using the online path planning methods. Some typical encounters situations are experimentally simulated. Simulation results show the effectiveness of the algorithm reasonable.
语种	中文
产权排序	1
页码	75页
分类号	TP242.3
内容类型	学位论文
源URL	[http://ir.sia.ac.cn/handle/173321/14777]
专题	沈阳自动化研究所_海洋信息技术装备中心
推荐引用方式 GB/T 7714	冷静. 面向实时避碰的无人水面机器人在线路径规划方法[D]. 中国科学院沈阳自动化研究所. 2014.