基于IGA的“类等效”建模在全自主足球机器人双闭环调速系统中的应用研究
|
摘要
全自主足球机器人是当今科学研究的热门领域之一。它集高科技和娱乐性于一身,是人工智能、机器人学、计算机视觉等领域,新理论、新方法的良好实验平台。
绝大部分控制系统的设计是在离线的情况下进行的。以什么样的模型在什么样的程度上,代替实际被控对象进行控制器的设计,是控制系统设计首先应当解决的关键问题之一。基于IGA的“类等效”建模方法,从被控对象的主要特征量出发,建立结构合理,参数精确的模型。这种方法极大的减小仿真模型和实际之间的差异,大大缩短仿真到实时控制之间的进程。
基于以上背景,本课题从全自主足球机器人的实际应用中引出我们需要建立模型的被控对象——双闭环调速系统。双闭环调速系统是构成直流电机驱动系统的典型方案,往往作为执行机构的重要组成部分,建立双闭环调速系统的模型具有广泛的实际意义。
本文先对直流电机双闭环调速系统的机理进行深入研究,抓住其主要的动态特征,应用“类等效”建模方法,把直流电机双闭环调速系统的模型,简化成一个简单的具有非线性特性的状态空间模型。进而提出了通过实测系统的转速响应,利用改进的遗传算法对该模型参数进行精确辨识的方法。
为检验方法的有效性,我们在General Bar全自主足球机器人上,建立了以MAXON电机、驱动器为主要硬件,及基于μC/OS-II嵌入式操作系统软件设计的实验平台。对General Bar全自主足球机器人的两套双闭环调速系统进行了“类等效”简化模型的建立。通过与常用模型的对比性实验,表明由此建模和参数辨识方法建立的模型,可以很好的替代实际调速系统,使整个机器人系统从定性分析转变为定量研究。
此外,我们以两套双闭环调速系统精确模型为基础,设计足球机器人运动控制仿真实验平台。通过在不同目标点下,仿真数据与里程计记录数据的比较可以证实,在仿真平台上设计的控制算法可以直接应用在实际中,加速了运动控制及上层策略系统的研究。
Soccer Robot is one of hot issue in technology research nowadays. It contains the high technology and the entertainment. Also, it is a right experiment platform for the new theories and methods in the area of artificial intelligence, robotic and the computer vision.
All most of the control system is designed off-line. The most important problem of the control system design which is must be solved firstly is how to found a model to subsititute the real controlled device. The "Quasi-Equivalent" modeling based on IGA, utilizing the main characters of the controlled device, can get a reasonable structural and parameters precise model. The method can reduce the difference of the model and the real object; shorten the transition form the simulation to the real-time control.
Based on the above background, double loop DC motor control system (DLM) is posed from the application of the autonomous soccer robot. The DLM is the typical approach to compose the DC motor drive system. It is very important part of the actuators. The modeling of the DLM has very broad meaning.
Firstly this paper made a deep study on the mechanism of the DLM. Then, using the method of characteristic analysis and“quasi-equivalent”modeling, the model of double loop DC motor control system has been reduced to a simple non-linear state space model. Further, based on the real rotate speed response character, the improved genetic algorithm has been indicated to accurately identify model parameters.
For verifying the validity of this method, we build an experiment platform on the General Bar autonomous soccer robot. The main hardware of the platform is MAXON motor and servoamplifier, while the software is design based on theμC/OS-II embed operating system. Moreover, the two sets of the DLM of the General Bar autonomous soccer robot are modeling by the“quasi-equivalent”method. Experiments of comparing this model with other common model proved the model which is established by“quasi-equivalent”method can substitute the real DLM. The simple model changes the whole system from qualitative analysis to quantitative research.
Furthermore, this paper design a motion control experiment platform based on the two sets of the DLM precise models. The comparing of the simulating data and the data recorded by the milemeter approve that the control algorithm designed on the simulate plafform can apply to the real situation diretly. The model speeds up the research of the motion control and the strategy system.
绝大部分控制系统的设计是在离线的情况下进行的。以什么样的模型在什么样的程度上,代替实际被控对象进行控制器的设计,是控制系统设计首先应当解决的关键问题之一。基于IGA的“类等效”建模方法,从被控对象的主要特征量出发,建立结构合理,参数精确的模型。这种方法极大的减小仿真模型和实际之间的差异,大大缩短仿真到实时控制之间的进程。
基于以上背景,本课题从全自主足球机器人的实际应用中引出我们需要建立模型的被控对象——双闭环调速系统。双闭环调速系统是构成直流电机驱动系统的典型方案,往往作为执行机构的重要组成部分,建立双闭环调速系统的模型具有广泛的实际意义。
本文先对直流电机双闭环调速系统的机理进行深入研究,抓住其主要的动态特征,应用“类等效”建模方法,把直流电机双闭环调速系统的模型,简化成一个简单的具有非线性特性的状态空间模型。进而提出了通过实测系统的转速响应,利用改进的遗传算法对该模型参数进行精确辨识的方法。
为检验方法的有效性,我们在General Bar全自主足球机器人上,建立了以MAXON电机、驱动器为主要硬件,及基于μC/OS-II嵌入式操作系统软件设计的实验平台。对General Bar全自主足球机器人的两套双闭环调速系统进行了“类等效”简化模型的建立。通过与常用模型的对比性实验,表明由此建模和参数辨识方法建立的模型,可以很好的替代实际调速系统,使整个机器人系统从定性分析转变为定量研究。
此外,我们以两套双闭环调速系统精确模型为基础,设计足球机器人运动控制仿真实验平台。通过在不同目标点下,仿真数据与里程计记录数据的比较可以证实,在仿真平台上设计的控制算法可以直接应用在实际中,加速了运动控制及上层策略系统的研究。
Soccer Robot is one of hot issue in technology research nowadays. It contains the high technology and the entertainment. Also, it is a right experiment platform for the new theories and methods in the area of artificial intelligence, robotic and the computer vision.
All most of the control system is designed off-line. The most important problem of the control system design which is must be solved firstly is how to found a model to subsititute the real controlled device. The "Quasi-Equivalent" modeling based on IGA, utilizing the main characters of the controlled device, can get a reasonable structural and parameters precise model. The method can reduce the difference of the model and the real object; shorten the transition form the simulation to the real-time control.
Based on the above background, double loop DC motor control system (DLM) is posed from the application of the autonomous soccer robot. The DLM is the typical approach to compose the DC motor drive system. It is very important part of the actuators. The modeling of the DLM has very broad meaning.
Firstly this paper made a deep study on the mechanism of the DLM. Then, using the method of characteristic analysis and“quasi-equivalent”modeling, the model of double loop DC motor control system has been reduced to a simple non-linear state space model. Further, based on the real rotate speed response character, the improved genetic algorithm has been indicated to accurately identify model parameters.
For verifying the validity of this method, we build an experiment platform on the General Bar autonomous soccer robot. The main hardware of the platform is MAXON motor and servoamplifier, while the software is design based on theμC/OS-II embed operating system. Moreover, the two sets of the DLM of the General Bar autonomous soccer robot are modeling by the“quasi-equivalent”method. Experiments of comparing this model with other common model proved the model which is established by“quasi-equivalent”method can substitute the real DLM. The simple model changes the whole system from qualitative analysis to quantitative research.
Furthermore, this paper design a motion control experiment platform based on the two sets of the DLM precise models. The comparing of the simulating data and the data recorded by the milemeter approve that the control algorithm designed on the simulate plafform can apply to the real situation diretly. The model speeds up the research of the motion control and the strategy system.
引文
[1] HirokiK, MinoruA, Yasuok. RoboCup: A challenge Problem for AI and Robotiec [M]. RoboCup-97: Robot Soccer World Cup I. Berlin: Springer, 1998.
[2]孙增圻.机器人足球综述[C].北京2003年全国机器人足球技术研讨会,北京:机器人技术与应用, 2003.9, 34-38.
[3]蔡自兴.机器人学的发展趋势和发展战略[J].机器人技术与应用. 2001, (4): 11-15.
[4]郭锐,吴敏,彭军,彭姣,曹卫华.一种新的多智能体Q学习算法[J].自动化学报, 2007, 33(4): 367-372.
[5]余存. RoboCup中型组足球机器人视觉系统的研究与设计[D].山东:山东大学, 2007.
[6]陈忠泽,林良明,颜国正.基于MAS(Multi-Agent System)的多机器人系统:协作多机器人学发展的一个重要方向[J].机器人, 2001.7, 23(4): 368-373.
[7]薛宏涛,叶媛媛,沈林成,常文森.多智能体系统体系结构及协调机制研究综述[J].机器人, 2001.1, 23(1): 85-90.
[8]陈卫东,席裕庚,顾冬雷.自主机器人的强化学习研究进展[J].机器人, 2001.7, 23(4): 379-384.
[9]陆军,徐莉,周小平.强化学习方法在移动机器人导航中的应用[J].哈尔滨工程大学学报, 2004.4, 25(2):176-179.
[10]段勇,徐心和.基于模糊神经网络的强化学习及其在机器人导航中的应用[J].控制与决策, 2007.5, 22(5): 525-534.
[11]刘娟,蔡自兴,涂春鸣.进化机器人学研究进展[J].控制理论与应用, 2002.8,19(4): 493-499.
[12] D.Floreano, J.Urzelai. Evolutionary robots with on-line self-organization and behavioral fitness [J]. Neural Networks,2000.5,13 (4): 431-443.
[13] Joanne H. Walker, Simon M.Garrett, Myra S. Wilson. The balance between initial training and lifelong adaptation in evolving robot controllers[J]. IEEE Transactions on Systems, Man, and Cybernetics, 2006.4,36 (2):423-432.
[14] A.L.Nelson, E. Grant, Developmental analysis in evolutionary robotics[C]. 2006 IEEE Mountain Workshop on Adaptive and Learning Systems, SMCals 2006. Logan, Utah, United States. 2006. Piscataway, NJ 08855-1331, United States: Institute of Electrical and Electronics Engineers Computer Society, 2006, 201-206.
[15]段清娟,王润孝.机器人学中应用的人工免疫原理模型与展望[J].机器人, 2005.9. 27(5): 475-480.
[16] Dipankar Dasgupta. An artificial immune system as a multi-agent decision support system[C]. Proceedings of the IEEE International Conference on Systems, Man and Cybernetics. San Diego, CA, USA.1998. Piscataway, NJ, USA: IEEE, 1998, 3816-3820.
[17]谢云.全自主机器人足球比赛系统的通信与多传感器信息融合技术[D].广东:广东工业大学, 2006.6.
[18]吴伟,刘兴刚,王忠实,徐心和.多传感器融合实现机器人精确定位[J].东北大学学报(自然科学版), 2007.2, 28(2): 161-164.
[19]王晓宇,闫继宏.两轮自平衡机器人多传感器数据融合方法研究[J].传感技术学报, 2007.3, 20(3): 668-672.
[20]陈伯时.电力拖动自动控制系统[M].北京:机械工业出版社,1997:48-55.
[21] Wu Weiguo, Chen Huitang, Woo Peng-Yung. Optimal motion planning for a wheeled mobile robot[C]. Proceedings-IEEE International Conference on Robotics and Automation, 1999, 1: 41-46.
[22]曹洋,方帅,徐心和.加速度约束条件下的非完整移动机器人运动控制[J].控制与决策, 2006, 21(2): 93-96.
[23]朱咏杰,付宜利,王树国,常疆.一种改进的移动机器人运动控制算法[J].哈尔滨工业大学学报. 2004, 36(1): 4-6.
[24]黄永志,陈卫东.两轮移动机器人运动控制系统的设计与实现[J].机器人, 2004, 26(1): 40-44.
[25]何艳丽,吴敏,曹卫华.自主机器人直线行进过程的数字PID控制[J].现代电子技术, 2005, (13): 58-60.
[26] L Huang. Speed Control of Differentially Driven Wheeled Mobile Robots-Model-Based adaptive approach [J]. Journal of Robotic Systems, 2005, 22(6): 323-332.
[27] Omar Al-Ayasrahl, T. Alukaidey, R. Salman, G. Pissanidis. Dual feed-back and feed-forward synchronized cross-coupled motion control for two-wheel mobile robot[C]. Proceedings of the 2005 IEEE International Workshop on Robotic Sensors: Robotic and Sensor Environments, ROSE 2005. IEEE. Canada, 2005: 55-60.
[28]杨承志,孙棣华,张长胜.系统辨识与自适应控制[M].重庆:重庆大学出版社, 2003.
[29]李祖枢.仿人智能控制理论与多级摆的摆起控制[M].人工智能:回顾与展望.北京:科学出版社, 2006.
[30]李祖枢,涂亚庆.仿人智能控制[M].北京:国防工业出版社, 2003.
[31]李祖枢.具有可调参数的模型降价新方法[J].自动化学报, 1982,(2):81-92.
[32]李祖枢,周其鉴,任伟.动态系统“类等效”模型测辨及多可调参数的Pade模型降阶[J].自动化学报, 1985,(1): 53-62.
[33]古建功,李祖枢,张华.三关节单杠体操机器人的动力学参数辨识[C].中国人工智能学会第11届全国学术年会论文集,北京:北京邮电大学出版社, 2005: 594-599.
[34]古建功.遗传算法在仿人智能控制中的应用[D].重庆:重庆大学, 2006.
[35] Jinn-Tsong Tsai. Tung-Kuan Liu. Jyh-Horng Chou. Hybrid Taguchi-Genetic Algorithm for Global Numerical Optimization [J]. IEEE trans. Evolutionary computation, Aug 2004, 8(4): 365-377.
[36] Y.W. Leung and Y.Wang. An orthogonal genetic algorithm with quantization for global numerical optimization [J]. IEEE Trans. Evol. Comput, Feb. 2001, 5(1), 41-53.
[37]刘和平. TMS320LF240X DSP结构,原理及应用[M].北京:北京航空航天大学出版社, 2002.
[38] Jean J. Labrosse.邵贝贝.嵌入式实时操作系统μC/OS-II [M].北京:北京航空航天大学出版社, 2003.
[39]于芳.基于动觉图式的仿人智能控制在移动机器人路径规划中的研究[D].重庆:重庆大学, 2007.
[40] Borenstein, J. and Feng, L. UMBmark– A method for Measuring, Comparing and Correcting Dead-reckoning Errors in Mobile Robots [M], USA: University of Michigan, 1994.
[41]王景川,陈卫东,曹其新.基于全景视觉与里程计的移动机器人自定位方法研究[J].机器人, 2005, 27(1) :41-45.
[2]孙增圻.机器人足球综述[C].北京2003年全国机器人足球技术研讨会,北京:机器人技术与应用, 2003.9, 34-38.
[3]蔡自兴.机器人学的发展趋势和发展战略[J].机器人技术与应用. 2001, (4): 11-15.
[4]郭锐,吴敏,彭军,彭姣,曹卫华.一种新的多智能体Q学习算法[J].自动化学报, 2007, 33(4): 367-372.
[5]余存. RoboCup中型组足球机器人视觉系统的研究与设计[D].山东:山东大学, 2007.
[6]陈忠泽,林良明,颜国正.基于MAS(Multi-Agent System)的多机器人系统:协作多机器人学发展的一个重要方向[J].机器人, 2001.7, 23(4): 368-373.
[7]薛宏涛,叶媛媛,沈林成,常文森.多智能体系统体系结构及协调机制研究综述[J].机器人, 2001.1, 23(1): 85-90.
[8]陈卫东,席裕庚,顾冬雷.自主机器人的强化学习研究进展[J].机器人, 2001.7, 23(4): 379-384.
[9]陆军,徐莉,周小平.强化学习方法在移动机器人导航中的应用[J].哈尔滨工程大学学报, 2004.4, 25(2):176-179.
[10]段勇,徐心和.基于模糊神经网络的强化学习及其在机器人导航中的应用[J].控制与决策, 2007.5, 22(5): 525-534.
[11]刘娟,蔡自兴,涂春鸣.进化机器人学研究进展[J].控制理论与应用, 2002.8,19(4): 493-499.
[12] D.Floreano, J.Urzelai. Evolutionary robots with on-line self-organization and behavioral fitness [J]. Neural Networks,2000.5,13 (4): 431-443.
[13] Joanne H. Walker, Simon M.Garrett, Myra S. Wilson. The balance between initial training and lifelong adaptation in evolving robot controllers[J]. IEEE Transactions on Systems, Man, and Cybernetics, 2006.4,36 (2):423-432.
[14] A.L.Nelson, E. Grant, Developmental analysis in evolutionary robotics[C]. 2006 IEEE Mountain Workshop on Adaptive and Learning Systems, SMCals 2006. Logan, Utah, United States. 2006. Piscataway, NJ 08855-1331, United States: Institute of Electrical and Electronics Engineers Computer Society, 2006, 201-206.
[15]段清娟,王润孝.机器人学中应用的人工免疫原理模型与展望[J].机器人, 2005.9. 27(5): 475-480.
[16] Dipankar Dasgupta. An artificial immune system as a multi-agent decision support system[C]. Proceedings of the IEEE International Conference on Systems, Man and Cybernetics. San Diego, CA, USA.1998. Piscataway, NJ, USA: IEEE, 1998, 3816-3820.
[17]谢云.全自主机器人足球比赛系统的通信与多传感器信息融合技术[D].广东:广东工业大学, 2006.6.
[18]吴伟,刘兴刚,王忠实,徐心和.多传感器融合实现机器人精确定位[J].东北大学学报(自然科学版), 2007.2, 28(2): 161-164.
[19]王晓宇,闫继宏.两轮自平衡机器人多传感器数据融合方法研究[J].传感技术学报, 2007.3, 20(3): 668-672.
[20]陈伯时.电力拖动自动控制系统[M].北京:机械工业出版社,1997:48-55.
[21] Wu Weiguo, Chen Huitang, Woo Peng-Yung. Optimal motion planning for a wheeled mobile robot[C]. Proceedings-IEEE International Conference on Robotics and Automation, 1999, 1: 41-46.
[22]曹洋,方帅,徐心和.加速度约束条件下的非完整移动机器人运动控制[J].控制与决策, 2006, 21(2): 93-96.
[23]朱咏杰,付宜利,王树国,常疆.一种改进的移动机器人运动控制算法[J].哈尔滨工业大学学报. 2004, 36(1): 4-6.
[24]黄永志,陈卫东.两轮移动机器人运动控制系统的设计与实现[J].机器人, 2004, 26(1): 40-44.
[25]何艳丽,吴敏,曹卫华.自主机器人直线行进过程的数字PID控制[J].现代电子技术, 2005, (13): 58-60.
[26] L Huang. Speed Control of Differentially Driven Wheeled Mobile Robots-Model-Based adaptive approach [J]. Journal of Robotic Systems, 2005, 22(6): 323-332.
[27] Omar Al-Ayasrahl, T. Alukaidey, R. Salman, G. Pissanidis. Dual feed-back and feed-forward synchronized cross-coupled motion control for two-wheel mobile robot[C]. Proceedings of the 2005 IEEE International Workshop on Robotic Sensors: Robotic and Sensor Environments, ROSE 2005. IEEE. Canada, 2005: 55-60.
[28]杨承志,孙棣华,张长胜.系统辨识与自适应控制[M].重庆:重庆大学出版社, 2003.
[29]李祖枢.仿人智能控制理论与多级摆的摆起控制[M].人工智能:回顾与展望.北京:科学出版社, 2006.
[30]李祖枢,涂亚庆.仿人智能控制[M].北京:国防工业出版社, 2003.
[31]李祖枢.具有可调参数的模型降价新方法[J].自动化学报, 1982,(2):81-92.
[32]李祖枢,周其鉴,任伟.动态系统“类等效”模型测辨及多可调参数的Pade模型降阶[J].自动化学报, 1985,(1): 53-62.
[33]古建功,李祖枢,张华.三关节单杠体操机器人的动力学参数辨识[C].中国人工智能学会第11届全国学术年会论文集,北京:北京邮电大学出版社, 2005: 594-599.
[34]古建功.遗传算法在仿人智能控制中的应用[D].重庆:重庆大学, 2006.
[35] Jinn-Tsong Tsai. Tung-Kuan Liu. Jyh-Horng Chou. Hybrid Taguchi-Genetic Algorithm for Global Numerical Optimization [J]. IEEE trans. Evolutionary computation, Aug 2004, 8(4): 365-377.
[36] Y.W. Leung and Y.Wang. An orthogonal genetic algorithm with quantization for global numerical optimization [J]. IEEE Trans. Evol. Comput, Feb. 2001, 5(1), 41-53.
[37]刘和平. TMS320LF240X DSP结构,原理及应用[M].北京:北京航空航天大学出版社, 2002.
[38] Jean J. Labrosse.邵贝贝.嵌入式实时操作系统μC/OS-II [M].北京:北京航空航天大学出版社, 2003.
[39]于芳.基于动觉图式的仿人智能控制在移动机器人路径规划中的研究[D].重庆:重庆大学, 2007.
[40] Borenstein, J. and Feng, L. UMBmark– A method for Measuring, Comparing and Correcting Dead-reckoning Errors in Mobile Robots [M], USA: University of Michigan, 1994.
[41]王景川,陈卫东,曹其新.基于全景视觉与里程计的移动机器人自定位方法研究[J].机器人, 2005, 27(1) :41-45.