SIA OpenIR  > 机器人学研究室
非完整系统鲁棒控制方法及在水面移动机器人中的应用
Alternative TitleRobust Control for Nonholonomic Systems and its Application on Unmanned Surface Vehicles
彭艳1,2
Department机器人学研究室
Thesis Advisor韩建达
ClassificationTP242.3
Keyword非完整约束 不可预知不确定性 主动建模 非线性控制 三体船型水面机器人(utv)
Call NumberTP242.3/P43/2009
Pages104页
Degree Discipline模式识别与智能系统
Degree Name博士
2009-05-26
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract近年来,随着机器人技术的发展,各种各样的移动机器人系统正在开始应用于工业、国防安全、公共安全、灾难救援、科学探测等领域;但是,遥控作业仍然是制约移动机器人广泛应用的一个核心问题。如何使机器人系统摆脱遥控的束缚而具有全自主行为能力,是目前机器人学领域重点研究的关键问题之一。自主环境适应能力是移动机器人最基本的自主能力之一。机器人系统需要面对不可预知的、动态的、存在时变扰动和噪声的工作环境,这通常会导致机器人动力学参数的变化,从而使传统的机器人控制技术很难得到满意的控制效果并达到自主环境适应的目的。同时,目前的移动机器人系统大都受非完整约束的影响,由于其约束的不可积性,使其控制与规划变得异常困难,因此研究移动机器人在存在不确定情况下的鲁棒控制问题,以及存在非完整约束条件下的镇定问题和规划问题具有非常重要的理论意义和实际应用价值。本文以沈阳自动化所与大连理工大学联合研制的三体船型水面移动机器人(UTV)这个典型的受不确定以及非完整约束影响的系统为背景,研究其在存在大量不确定因素及外界环境干扰条件下的鲁棒控制问题和其在受制于非完整约束情况下的镇定和运动规划问题,其中前者主要针对系统在受风、流、浪等外界干扰下的鲁棒控制问题;后者旨在解决满足非完整约束条件下的非线性系统规划与控制问题。本论文的具体内容安排如下:第1章,简单介绍了非完整系统的概念,归纳出非完整系统在控制上的两方面核心问题,即:不确定性及扰动的抑制问题和运动规划与控制问题。对解决这两类问题的现有方法进行了深入分析和综述,包括动态模型的在线估计共性方法、非完整约束系统运动控制方法、以及优化与预测控制方法,同时,介绍了本文作为仿真与实验对象的三体船型水面平台的发展概况。针对现有方法存在的问题有针对性地提出了本论文的研究内容。第2章,首先简要介绍了机器人学国家重点实验室与大连理工大学联合研制的三体船型水面移动机器人试验平台,作为本论文后续仿真与实验研究的对象,介绍了该平台的动力系统与控制系统。从运动控制的角度出发,综合考虑推力、水动力等对整个系统的动力学进行了分析,建立了完整的6自由度动力学模型以及水平面运动的简化模型。在此基础上,分析了三体船型水面移动机器人的非完整约束特性和能控性,为后续方法研究奠定了基础。第3章,提出了一种基于主动建模的鲁棒控制设计方法。对于包含难以准确建模动力学以及外来扰动等不确定性的系统,可以将其动力学模型合理简化,而将所有不确定因素以模型差的形式引入到系统中;将引入的模型差与系统原有状态组合成增广状态,构造出联合估计模型;再利用在线估计方法实时估计这个模型差;同时,将估计出的模型差利用某种控制策略反馈到控制量中,达到增强系统鲁棒性的目的。针对三体船型水面移动机器人,在理论上分析了上述方法的稳定性,并在实际三体船型水面机器人上开展了实验比较研究。实验结果表明此方法对系统中存在的各种不确定因素具有很好的抑制作用。第4章,重点研究非线性系统在线估计以及将在线与控制相结合的问题。在简要介绍可以用于非线性估计的UKF方法的基础上,针对其性能严重依赖噪声先验知识的问题,提出了一种基于主从结构的自适应滤波器设计方法。同时针对三体船型水面机器人非线性系统模型中存在的不确定性因素,提出了一种基于自适应UKF的指数跟踪控制器设计方法。仿真结果验证了自适应UKF算法的性能提高,同时验证了基于在线估计的指数跟踪控制器对时变参数与扰动的抑制作用。第5章,研究了非完整约束系统的镇定控制问题。提出了一种高维流形嵌入法;通过把二阶非完整约束系统嵌入到高维流形中,将在原空间中镇定到一个点的问题转化成在扩展的状态空间中镇定到一个子流形的问题;从而,可以将原空间不可连续反馈镇定的控制问题转化为高维流形中可连续反馈镇定控制问题,理论上证明了该方法的正确性和完备性。针对三体船型水面移动机器人的仿真结果验证了该方法的正确性。第6章,针对非完整约束对移动机器人可执行轨迹的制约问题,提出了跟踪控制Lyapunov函数的概念,并将其引入到非完整系统地运动规划中,用跟踪控制Lyapunov函数来保证规划算法的收敛性和可执行性。该方法将非线性预测控制与基于跟踪控制Lyapunov函数的广义逐点最小范数控制相结合,得到了一种在保证闭环稳定性前提下的运动规划算法;理论上证明了该方法的正确性,并针对三体船型水面移动机器人动力学进行仿真验证。
Other AbstractWith the development of relative techniques, robots, especially various mobile robots, are becoming more and more applicable in the fields such as industry, national defense, public security, disaster rescue, and scientific exploration. However, most of the mobile robots still have to be controlled by human operator, namely, by the tele-operation. The limitation in autonomy is one of the most critical problems, which restrains the extensive applications of mobile robots. Therefore, control for autonomy has been becoming an active research direction in robotics. Environment adaptability is the fundamental performance of an autonomous mobile robot, which usually needs to work in dynamic environments with unpredictable disturbances and noises. These ‘external’ uncertainties will further introduce internal variations into the robot dynamics. Both the internal and external uncertainties will degrade the control performance and need compensation techniques to autonomously handling. At the same time, most of mobile robots also suffer from nonholonomic constraint, which makes the robust control and motion planning even more difficult. This thesis focuses on the control methodology that can improve the environment tolerance performance of a mobile robot under nonholonomic constraint, while also suffering unpredictable uncertainties. With respect to unknown and time-varying uncertainties, a valid active-model based control scheme is proposed, and the corresponding nonlinear estimation techniques, as well as the robust control algorithms capable of integrating the estimated uncertainties, are studied in details. A new control design method, which is based on the proposed Manifold Transformation, and a model predictive control type path planner, which takes into account the nonholonomic constraint, are also proposed to stabilize the nonholonomic system and generate its executable trajectory respectively. The validity of all the proposed methods is theoretically proved, and the performances are verified and demonstrated by simulations or experiments on the Unmanned Trimaran Vehicle (UTV) collaboratively designed by Shenyang Institute of Automation and Dalian University of Technology. The main contents of this thesis are organized as follows: In Chapter 1, the concept of nonholonomic system is simply introduced, and the two critical problems involved, i.e., disturbance rejection and stabilization/motion planning, are summarized. Available methods working on these two problems are analyzed, and therefore the motivations and research topics are explained with respect to the open problems. In Chapter 2, the UTV testbed is described. The 6-DOF dynamic equations, which can describe the complicated motion of the vehicle, are established. The controllability as well as the nonholonomic constraint are further analyzed. As a typical nonholonomic system, all the theoretical analyses, simulations and experiments in the following chapters, are conducted with respect to the dynamics or on the UTV testbed. In Chapter 3, an active model based control scheme is proposed. The complicated dynamics are simplified and the ignored elements as well as all the unpredictable uncertainties are summarized into a variable named model error. A reference model, which includes this model error as an extended state, is presented and a joint estimator is proposed to actively estimate the model error. The estimated results are further fed back into controller for model error rejection. This proposed scheme is verified by real experiments of the UTV heading control. Chapter 4 studies how the active model based control can be applied to nonlinear systems. A master-slave adaptive UKF (AUKF) is proposed to handle nonlinear estimation. The adaptive mechanism intends to reduce the dependence of normal UKF on the priori knowledge of the noise distribution, and increase the robustness while it is used in high dynamic environment. With respect to the dynamics of UTV, an AUKF based exponential tracking controller is also designed. Simulations demonstrate the performance improvement of the proposed method. In Chapter 5, a manifold transformation method is proposed in order to design the continuous stabilization control of nonholonomic systems. By embedding the second order nonholonomic system into a higher order manifold, the point stabilizing problem can be transferred to submanifold stabilizing problem in the new manifold, where the continuous feedback control law can be easily designed. The validity of the proposed method is theoretically proved and its performance is also demonstrated by the simulations conducted on the UTV dynamics. Chapter 6 studies the trajectory generation method that can guarantee the generated trajectory is executable under nonholonomic constraint. A tracking control Lyapunov function (TCLF) is proposed, and integrated into the so called ‘generalized pointwise min-norm controller (GPMNC)’, which is a kind of nonlinear model predictive control (NMPC) schemes. The TCLF is to guarantee that the trajectory generation is convergent and executable under nonholonomic constraint, and the GPMNC will help to achieve the goal of optimal planning. This new algorithm is proved theoretically, and simulation results show its effectiveness.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/502
Collection机器人学研究室
Affiliation1.中国科学院沈阳自动化研究所
2.中国科学院研究生院
Recommended Citation
GB/T 7714
彭艳. 非完整系统鲁棒控制方法及在水面移动机器人中的应用[D]. 沈阳. 中国科学院沈阳自动化研究所,2009.
Files in This Item:
File Name/Size DocType Version Access License
10001_20061801472803(12458KB) 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
[彭艳]'s Articles
Baidu academic
Similar articles in Baidu academic
[彭艳]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[彭艳]'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.