FastSLAM2.0控制法则与PGR导航法则的结合研究及其仿真
摘要
通过对现代国内外机器人现状和机器人相关的基本理论的介绍,我们了解到了现在机器人中的热点研究话题一定位问题。而定位与制图问题恰恰是移动机器人应用中的的重要问题之一。为了更好地解决这个问题,本文选择了FastSLAM,FastSLAM是将SLAM分解为机器人定位和特征标志的位置估计两个过程。粒子滤波器中的每个粒子代表机器人的一条可能运动路径,利用观测信息计算每个粒子的权重,以评价每条路径的好坏,从而可以提高其估计的准确性和可靠性。
在机器人运动过程中的另一个重要的问题,就是导航控制(路径规划)。在路径跟踪控制问题方面,由于在笛卡尔坐标系下不能应用反馈线性化或光滑定常反馈的控制器设计方法渐近镇定系统。本文提出了尝试用非线性预测控制镇定有控制量约束和运动学约束的移动机器人,提出了一种新的导航算法--PGR算法,并且详细介绍了该算法的整个数学的推导过程。
在以上理论的研究基础上,鉴于Matlab价格昂贵且使用上受到限制,在国内售后服务和技术支持也有限,而且本文的仿真的环境也不是很复杂,因此,本文选择了一款可以和Matlab相匹敌的完全免费的自由软件scilab进行仿真。在仿真过程中,首先编写了内部传感器和外部传感器的数据采集文件,用来模拟机器人在移动过程中传感器反馈回来的信息。同时也将定位算法与导航算法分别用scilab语言编写成程序文件。最后,设计了机器人整个运动过程的控制流程图,然后根据流程图和这些已建立好的数据信息对移动机器人的运动环境进行大量仿真。通过完成该设计,对现实中的移动机器人的实现也很容易,这样根据这个控制实现的可能性,跟人驾驶的车相似的车辆的运动控制就可以用该方法实现,对前方没有目标地点的情况动作也可能实现了。
By introducing the basic theory on modern robot ,we know the topic is the location of robot. The robot location and mapping is the important issues in the application of robot. To solve this problem better, this paper chose FastSLAM, FastSLAM made SLAM into two processes that robot estimated location and estimated the features' positions. Each particle in the particle filter indicates a robot movement path. We can use the observation information to calculate the weight of each particle and get the appraise of every path ,then we can improve both the veracity and the reliability.
In the course of the robot application , another important issue is navigation control (path planning). In the path tracking control issues, as the Cartesian coordinate system of feedback can not be linear or smooth steady feedback controller design methods asymptotic stabilization system. In this paper, we try to use non-linear predictive control calm and restraint in control of the mobile robot kinematics constraints. So we proposed a new navigation algorithm - PGR algorithm.
However, Matlab has a high price and have no enough support on service and technique. Basing on above reasons , this article use the Scilab to do large numbers of experiments. During the simulation process, we collect the data prepared by the internal sensor and external sensor, the data can be used to simulate the movement of a robot by the sensor feedback information. SLAM algorithm and navigation algorithm were written into programm files by scilab. Finally, we get the process flow chart of a mobile robot, then build the simulation environment of mobile robot by all data. By the complete design, the mobile robot can easily move in the reality .To control the vehicles movement similar to the vehicles with people driving, thus the movement of no target may also be true in reality.
在机器人运动过程中的另一个重要的问题,就是导航控制(路径规划)。在路径跟踪控制问题方面,由于在笛卡尔坐标系下不能应用反馈线性化或光滑定常反馈的控制器设计方法渐近镇定系统。本文提出了尝试用非线性预测控制镇定有控制量约束和运动学约束的移动机器人,提出了一种新的导航算法--PGR算法,并且详细介绍了该算法的整个数学的推导过程。
在以上理论的研究基础上,鉴于Matlab价格昂贵且使用上受到限制,在国内售后服务和技术支持也有限,而且本文的仿真的环境也不是很复杂,因此,本文选择了一款可以和Matlab相匹敌的完全免费的自由软件scilab进行仿真。在仿真过程中,首先编写了内部传感器和外部传感器的数据采集文件,用来模拟机器人在移动过程中传感器反馈回来的信息。同时也将定位算法与导航算法分别用scilab语言编写成程序文件。最后,设计了机器人整个运动过程的控制流程图,然后根据流程图和这些已建立好的数据信息对移动机器人的运动环境进行大量仿真。通过完成该设计,对现实中的移动机器人的实现也很容易,这样根据这个控制实现的可能性,跟人驾驶的车相似的车辆的运动控制就可以用该方法实现,对前方没有目标地点的情况动作也可能实现了。
By introducing the basic theory on modern robot ,we know the topic is the location of robot. The robot location and mapping is the important issues in the application of robot. To solve this problem better, this paper chose FastSLAM, FastSLAM made SLAM into two processes that robot estimated location and estimated the features' positions. Each particle in the particle filter indicates a robot movement path. We can use the observation information to calculate the weight of each particle and get the appraise of every path ,then we can improve both the veracity and the reliability.
In the course of the robot application , another important issue is navigation control (path planning). In the path tracking control issues, as the Cartesian coordinate system of feedback can not be linear or smooth steady feedback controller design methods asymptotic stabilization system. In this paper, we try to use non-linear predictive control calm and restraint in control of the mobile robot kinematics constraints. So we proposed a new navigation algorithm - PGR algorithm.
However, Matlab has a high price and have no enough support on service and technique. Basing on above reasons , this article use the Scilab to do large numbers of experiments. During the simulation process, we collect the data prepared by the internal sensor and external sensor, the data can be used to simulate the movement of a robot by the sensor feedback information. SLAM algorithm and navigation algorithm were written into programm files by scilab. Finally, we get the process flow chart of a mobile robot, then build the simulation environment of mobile robot by all data. By the complete design, the mobile robot can easily move in the reality .To control the vehicles movement similar to the vehicles with people driving, thus the movement of no target may also be true in reality.
引文
[1]郭戈,胡征峰,董江辉。微计算机信息。北京:微计算机信息出版社,2003,8:10-11.
[2]蔡自兴等。控制与决策,沈阳:控制与决策编辑部2002,17(4):385-390
[3]迟健男,徐心和。机器人。沈阳:《机器人》编辑部,2004,26(1):92-96.
[4]吕丹,戴学丰,刘树东.微机算机信息.北京:微计算机信息出版社,1989:40-48
[5]张恒,樊晓平.机器人.沈阳:《机器人》编辑部.2006.5:285-290
[6]项志宇,刘济林。仪器仪表学报。北京:仪器仪表学会,2003。8(24):391-394
[7]胡劲草。传动技术。上海:上海交通大学,2006.12(20):14-18
[8]鞠纯纯,何波,刘保龙,王永清。系统仿真学报。北京:北京大学出版 2007。8:3715-3718
[9]厉茂海,洪炳镕,罗荣华。哈尔滨工业大学学报。哈尔滨:哈尔滨工业大学编辑部,2004.7
[10]皮旷怡,马孜,徐慧朴。控制工程。沈阳:《控制工程》编辑部2007。5:162-165
[11]罗荣华,洪炳铬。机器人。沈阳:《机器人》编辑部,2006。8:84-86
[12]陈东岳,张立明。控制理论与应用。广州:控制理论与应用编辑部,2006.10:679-686
[13]张柱华,方华京。计算机与数学工程。湖北:《计算机与数字工程》编辑部,2005.7:53-55
[14]Thruns S,Fox D,Burgard W。A probabilistic approach to concurrent mapping and localization for mobile robots[J]。 Machine Learning,1998,31(1-3):29- 53.
[15]kiu Y,Thrun S。Results for outdoor- SLAM using sparse extended information filters [A]。Proc of the IEEE Int Conf on Robotics&Automation[C]。Taipei:IEEE Press,2003:1227-1233.
[16]Montemerlo M,Thrun So Simultaneous localization and mapping with unknown data association using FastSLAM[A]。Proc of the IEEE Int Conf on Robotics &Automation[C]。Taipei:IEEE Press,2003:1985-1991.
[17]。A。Elfes。Using occupancy grids for mobile robot perception and navigation,IEEE Computer,1989,22(6):46-57。
[18]。Chong K S,Kleem an Lindsay。Mobile robot map building from an advanced sonar array and accurate odometry,International Journal of Robotics Research,1999,18(1):20-36.
[19]。Thrun S,Bucken A。Integraing grid-based and top,logical maps for mobile robot navigation。In Proceeding Twelfth NCAI,AAAI,1996.
[20]Ding Z,Leung H,nong LDecoupling jolnt prnbabilistic data ass.cation algorithm for multiple target tracking叨.IEE Proceedings Radar Sonar Navigation,1999,142(1):11-17
[21]C。CrowandF。L。Manning.DieffrentialGpScontrolofstarearZ。Navigation,Jounralofthen InstituteofNavigation,39(4):383-405,Winter1992-93.
[22]Kevin Murphy.Bayesian Map Learning in Dynamic Environments[J].Nil' s(S0944-8535),1999,5(7):82-88
[23]S Thrun,Y Liu。Simultaneous Localization and Mapping with Sparse Extended Information Filters[J].International Journal of Robotics Research(S0278-364 9),2004,7(11):67-75.
[24]#12
[25]#12
[26]#12
[27]#12
[28]李庆扬、关治、白峰杉,数值计算原理,北京:清华大学出版社,2000
[29]李涛、贺勇军等,Matlab工具箱应用指南-应用数学篇,北京:电子工业出版社,2000
[30]胡包钢,赵星,康盂珍·Scilab教程,北京:清华大学出版社,2003
[31]周超、白峰杉,《数值计算与计算机应用》,北京:《数值计算与计算机应用》编辑部 2002,6(2):144-152.
[2]蔡自兴等。控制与决策,沈阳:控制与决策编辑部2002,17(4):385-390
[3]迟健男,徐心和。机器人。沈阳:《机器人》编辑部,2004,26(1):92-96.
[4]吕丹,戴学丰,刘树东.微机算机信息.北京:微计算机信息出版社,1989:40-48
[5]张恒,樊晓平.机器人.沈阳:《机器人》编辑部.2006.5:285-290
[6]项志宇,刘济林。仪器仪表学报。北京:仪器仪表学会,2003。8(24):391-394
[7]胡劲草。传动技术。上海:上海交通大学,2006.12(20):14-18
[8]鞠纯纯,何波,刘保龙,王永清。系统仿真学报。北京:北京大学出版 2007。8:3715-3718
[9]厉茂海,洪炳镕,罗荣华。哈尔滨工业大学学报。哈尔滨:哈尔滨工业大学编辑部,2004.7
[10]皮旷怡,马孜,徐慧朴。控制工程。沈阳:《控制工程》编辑部2007。5:162-165
[11]罗荣华,洪炳铬。机器人。沈阳:《机器人》编辑部,2006。8:84-86
[12]陈东岳,张立明。控制理论与应用。广州:控制理论与应用编辑部,2006.10:679-686
[13]张柱华,方华京。计算机与数学工程。湖北:《计算机与数字工程》编辑部,2005.7:53-55
[14]Thruns S,Fox D,Burgard W。A probabilistic approach to concurrent mapping and localization for mobile robots[J]。 Machine Learning,1998,31(1-3):29- 53.
[15]kiu Y,Thrun S。Results for outdoor- SLAM using sparse extended information filters [A]。Proc of the IEEE Int Conf on Robotics&Automation[C]。Taipei:IEEE Press,2003:1227-1233.
[16]Montemerlo M,Thrun So Simultaneous localization and mapping with unknown data association using FastSLAM[A]。Proc of the IEEE Int Conf on Robotics &Automation[C]。Taipei:IEEE Press,2003:1985-1991.
[17]。A。Elfes。Using occupancy grids for mobile robot perception and navigation,IEEE Computer,1989,22(6):46-57。
[18]。Chong K S,Kleem an Lindsay。Mobile robot map building from an advanced sonar array and accurate odometry,International Journal of Robotics Research,1999,18(1):20-36.
[19]。Thrun S,Bucken A。Integraing grid-based and top,logical maps for mobile robot navigation。In Proceeding Twelfth NCAI,AAAI,1996.
[20]Ding Z,Leung H,nong LDecoupling jolnt prnbabilistic data ass.cation algorithm for multiple target tracking叨.IEE Proceedings Radar Sonar Navigation,1999,142(1):11-17
[21]C。CrowandF。L。Manning.DieffrentialGpScontrolofstarearZ。Navigation,Jounralofthen InstituteofNavigation,39(4):383-405,Winter1992-93.
[22]Kevin Murphy.Bayesian Map Learning in Dynamic Environments[J].Nil' s(S0944-8535),1999,5(7):82-88
[23]S Thrun,Y Liu。Simultaneous Localization and Mapping with Sparse Extended Information Filters[J].International Journal of Robotics Research(S0278-364 9),2004,7(11):67-75.
[24]#12
[25]#12
[26]#12
[27]#12
[28]李庆扬、关治、白峰杉,数值计算原理,北京:清华大学出版社,2000
[29]李涛、贺勇军等,Matlab工具箱应用指南-应用数学篇,北京:电子工业出版社,2000
[30]胡包钢,赵星,康盂珍·Scilab教程,北京:清华大学出版社,2003
[31]周超、白峰杉,《数值计算与计算机应用》,北京:《数值计算与计算机应用》编辑部 2002,6(2):144-152.