题名互联网遥操作机器人系统的预测控制研究
作者陈丹
学位类别博士
答辩日期2010-12-12
授予单位中国科学院沈阳自动化研究所
授予地点中国科学院沈阳自动化研究所
导师席宁
关键词Internet 机器人遥操作 预测控制 事件控制
其他题名Research on Predictive Control of Internet-based Telerobotic System
学位专业模式识别与智能系统
中文摘要随着Internet技术的飞速发展,Internet已经成为人类动作行为的载体,从而可实现人类操作功能的延伸。 近年来,众多学者将Internet与机器人相连,推动了信息领域与机器人领域的融合。 基于Internet的网络遥操作成为目前的热门研究课题之一。 基于Internet的遥操作是从遥操作基础上发展起来的,遥操作方面的许多理论、方法对基于Internet的遥操作系统依然适用或者具有一定的启迪作用,然而Internet所特有的可变时延、数据丢包和乱序以及多通道传输等特点使得基于Internet遥操作系统的控制控制器设计和系统稳定性分析变得更为复杂和困难。 本文在分析了现有的遥操作技术和理论的基础上,系统地分析互联网控制系统的特性,网络通信的随机时延的统计特征,及其对控制系统的稳定性、可控性等性能的影响,提出了采用了事件驱动与预测控制方法相结合的方法,研究网络遥操作系统的建模与控制问题: 1. 在分析了网络机器人遥操作系统若干关键问题之后,结合以往的基于网络的遥操作机器人系统,提出的基于事件预测的网络遥操作机器人系统结构模型。 2. 测试了Internet中端到端的网络随机时延,并分析了Internet中端到端的可变时延对系统稳定性和透明性的影响,指出对于时变时延的统计特征是符合平移gamma分布,根据平移gamma分布参数变化情况,以及时延数据的相关性,提出了基于可变滑动窗口的稀疏多元线性回归算法的回路时延短期预测算法。 3. 围绕基于事件预测的网络遥操作系统结构模型,在分析基于事件的规划与控制策略的基础上,提出了基于网络的遥操作机器人系统主端的事件预测优化策略,并设计了路径管理器。 通过稳定性分析给出系统的稳定性条件:预测的事件是时间的非递减函数,并且要大于或等于当前系统事件。 4. 研究网络数据丢包条件下的机器人控制问题。 提出了在主端的预测事件的基础上,从端设计一个广义预测控制器,进行多步预测,当从端在一个采样周期内没有收到主端发出的指令时,则用广义预测控制器的预测输出作为当前的控制指令控制机器人。 5. 搭建具有力反馈功能的、基于客户机/服务器模式的、具有三层分布式控制结构的移动机器人网络遥操作系统实验平台,并为系统设计了人机交互接口。 通过所搭建的实验系统对理论和方法的研究进行了实验验证,实验结果证明了所提出的方法是有效可行的,对于建立性能良好的基于网络的遥操作机器人系统具有现实意义。
英文摘要With the rapid development of the Internet technology, the Internet has been a carrier of human’s behavior so as to extend human’s actions and intelligence to remote locations. In recent years, the Internet and robots are closely connected by many scholars so as to promote the conbination of the information field and robotic field. The Internet-based teleoperation has been received much attention.     The Internet-based telecontrol is developed from teleoperation which theories and methods can be used or enlinghtenment in the Internet-based teleoperation.Due to the ubiquitous nature of the Internet, such as variable timedelay, packet loss and disorder, it is very complicated and hard to design the the controller and analyze the stability of the Internet-based teleoperation system. After analyzing existing teleoperation technology and theoies, we analysed the characteristics of the Internet-based telecontrol system, and statistical characteristics of the Internet variable time-delay. It is found that how these characteristics affect the controllability, observability and stablility of the teleoperation system. We proposed the pedictive control strategy combined the event-based plan method to study the modelling and controlling the Internet-based teleoperation system: 1. After analysing some key questions of the Internet-based teleoperation system, and according to the previous Internet-based teleoperation system, we proposed a new Internet-based teleoperation system structure which used event-prediction method. 2. We test the end-to-end random round trip time-delay(RTT) in Internet and analyze the variable RTT how to affect the stability and transparency of the teleoperation system. And then, we conclude that the probability density of daily RTT would follow shifted Gamma distributions. According to the parameters variation of these shifted Gamma distributions and the long-range correlations of the RTT, we propose a sparse multivariate linear regressive (SMLR) model with variable slipping window for the one-step RTT prediction. 3. As the matter of the proposed event-prediction based teleoperation system structure, we propose the event-based predictive optimizing scheme on the basis of event-based planning and control strategy, and design a Path Governor at master site. Through stability analysis, we can acquire the following stability condition: If the predictive event is greater than or equal to the system event, and the system event is a nondecreasing function of time, the event predictive control strategy can make the telerobotics stable. 4. Based on studying the robotic control problem under the Internet packet loss conditions, we propose the Multi-step-ahead Generalized Predictive Controller(GPC) at slave site on the basis of the predictive event at master site. When the robot didn’t receive the commands the master sent in a sample period, the predictive item out of GPC would be used to control the robot. 5. The networked mobile robot system based on the Client/Server structure is constructed, which has the force feedback function and 3-layer distributed control structure. We also design the man-machine interface.      The experimental results on the networked mobile robotic system show that the proposed model and control method is effective, feasible and significant to the establishment of good networked robotic systems.
语种中文
产权排序1
公开日期2012-07-27
内容类型学位论文
源URL[http://ir.sia.ac.cn/handle/173321/9380]  
专题沈阳自动化研究所_机器人学研究室
推荐引用方式
GB/T 7714
陈丹. 互联网遥操作机器人系统的预测控制研究[D]. 中国科学院沈阳自动化研究所. 中国科学院沈阳自动化研究所. 2010.
个性服务
查看访问统计
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。


©版权所有 ©2017 CSpace - Powered by CSpace