RoboCup3D仿真系统仿人机器人定位与角色研究
|
摘要
机器人技术是一项综合性的应用技术,高度交叉了包括人工智能、机械、电子以及自动化控制在内的多门学科,是当代最智能的自动化应用之一。仿人机器人作为真正字面意义上或狭义上的“机器人”更是机器人技术中的一个热点研究方向,代表了机器人学的尖端研究水平。
在实际应用中往往有多个机器人同时出现,从而产生的分布式多智能体系统(Multi Agent System)的问题成为我们研究的重点。多智能体之间的协作问题中每个机器人都应有自己的职责,简单有效的对机器人进行动态角色分配才能更好的适应环境的动态变化。而角色选择的一个重要依据就是机器人所处的位置,进而需要能够用受限的视觉在含有噪声的动态环境中准确定位。
本文主要研究基于RoboCup3D仿真平台,具体工作如下：
1、阐述机器人运动学基础,根据仿人机器人结构对其进行运动学建模,通过正运动学算法给出从机器人关节角度到腿部位姿的求解,同时应用逆运动学解析法将机器人的步行问题转换为关节角度形式。此部分的内容是比赛的基础,也是后面所要研究的内容的基本保证。
2、机器人在足球仿真比赛中根据视觉来进行自定位和对其他物体的定位,而视觉模型采用的是局部视觉,为此给出不同情况下的对机器人的定位方法。并针对视觉信息的噪声干扰提出一种基于卡尔曼滤波的改进定位方法,对观测到含有噪声视觉信息去除噪声,从而提高了定位的准确度。
3、机器人足球仿真3D已从最初的3对3比赛变成了11对11的比赛,因此,如何更加有效进行球员角色分配和阵型的选择显得尤为重要。此部分给出了球队的阵型排布,为每个球员角色分配专门的职责,并提出一种以动态规划为基础的动态角色分配策略,使得每个球员根据比赛场上的动态变化情况及时做出角色切换,满足比赛的要求。
本文将上述工作应用到安徽大学RoboCup3D仿真球队Dream Wing3D中,提高了球队的攻防能力,更好的完成了比赛。
Robot technology is a comprehensive application technology, highly involved multiple disciplines like artificial intelligence, machinery, electronics and automatic control, and is the most intelligent automation applications today. Humanoid robot as a real literal sense or in the narrow sense "robot" is a hot research direction that stands for the tip of robotics research level.
In practical applications, multiple robots often appeared at the same time, resulting in a distributed multi agent system which has become our research hotspot. Each robot should have their own responsibilities in multi agent system collaboration, and do dynamic role assignment simple and effective for robot to adapt to the dynamic changes of the environment better. And the role selection is gravely based on the robot's location; therefore the most important problem is accurate localization in limited visual uncertain, noisy and dynamic environment.
Based on the simulation platform of RoboCup3D, The specific work this paper studies as follows:
1. Elaborate the basis of robot kinematics, do the kinematical modeling based on humanoid robot structure, solve robot joint angle to the leg pose using forward kinematics algorithm, and use inverse kinematics analytic method to converts robot walking problem to joint angles problem. This part is the basic of Robot competition, and the implementation of following content will based on this part.
2. In RoboCup simulation game, robots get the positions of their own and the other objects according to vision, but this vision model is a restricted vision model. Thus according to different number of conditions there are different robot localization methods. Aim to visual information noise, proposed an improved localization method based on the Kalman Filter. A prediction method of the position of the robot is given by the analysis of the motion model simulation game, combined with the Kalman Filter theory to get rid of the noise information observed to improve the localization accuracy.
After years of development, RoboCup3D simulation game has been more and more close to the human football game form the initial3vs3game into11vs11. Therefore, how to assign the players role and of choose formation more effectively is very important. This part we shows an arrangement of the team's formation, and assigns special duties to each player, and put forward a dynamic role assignment policy based on dynamic programming, make each player switch their roles timely according to the dynamic changes of the playing condition to meet the requirements of the game.
The above work applied to the Anhui University of RoboCup3D simulation team called DreamWing3D, and improved the team's offensive and defensive capabilities, better completed the competition.
在实际应用中往往有多个机器人同时出现,从而产生的分布式多智能体系统(Multi Agent System)的问题成为我们研究的重点。多智能体之间的协作问题中每个机器人都应有自己的职责,简单有效的对机器人进行动态角色分配才能更好的适应环境的动态变化。而角色选择的一个重要依据就是机器人所处的位置,进而需要能够用受限的视觉在含有噪声的动态环境中准确定位。
本文主要研究基于RoboCup3D仿真平台,具体工作如下：
1、阐述机器人运动学基础,根据仿人机器人结构对其进行运动学建模,通过正运动学算法给出从机器人关节角度到腿部位姿的求解,同时应用逆运动学解析法将机器人的步行问题转换为关节角度形式。此部分的内容是比赛的基础,也是后面所要研究的内容的基本保证。
2、机器人在足球仿真比赛中根据视觉来进行自定位和对其他物体的定位,而视觉模型采用的是局部视觉,为此给出不同情况下的对机器人的定位方法。并针对视觉信息的噪声干扰提出一种基于卡尔曼滤波的改进定位方法,对观测到含有噪声视觉信息去除噪声,从而提高了定位的准确度。
3、机器人足球仿真3D已从最初的3对3比赛变成了11对11的比赛,因此,如何更加有效进行球员角色分配和阵型的选择显得尤为重要。此部分给出了球队的阵型排布,为每个球员角色分配专门的职责,并提出一种以动态规划为基础的动态角色分配策略,使得每个球员根据比赛场上的动态变化情况及时做出角色切换,满足比赛的要求。
本文将上述工作应用到安徽大学RoboCup3D仿真球队Dream Wing3D中,提高了球队的攻防能力,更好的完成了比赛。
Robot technology is a comprehensive application technology, highly involved multiple disciplines like artificial intelligence, machinery, electronics and automatic control, and is the most intelligent automation applications today. Humanoid robot as a real literal sense or in the narrow sense "robot" is a hot research direction that stands for the tip of robotics research level.
In practical applications, multiple robots often appeared at the same time, resulting in a distributed multi agent system which has become our research hotspot. Each robot should have their own responsibilities in multi agent system collaboration, and do dynamic role assignment simple and effective for robot to adapt to the dynamic changes of the environment better. And the role selection is gravely based on the robot's location; therefore the most important problem is accurate localization in limited visual uncertain, noisy and dynamic environment.
Based on the simulation platform of RoboCup3D, The specific work this paper studies as follows:
1. Elaborate the basis of robot kinematics, do the kinematical modeling based on humanoid robot structure, solve robot joint angle to the leg pose using forward kinematics algorithm, and use inverse kinematics analytic method to converts robot walking problem to joint angles problem. This part is the basic of Robot competition, and the implementation of following content will based on this part.
2. In RoboCup simulation game, robots get the positions of their own and the other objects according to vision, but this vision model is a restricted vision model. Thus according to different number of conditions there are different robot localization methods. Aim to visual information noise, proposed an improved localization method based on the Kalman Filter. A prediction method of the position of the robot is given by the analysis of the motion model simulation game, combined with the Kalman Filter theory to get rid of the noise information observed to improve the localization accuracy.
After years of development, RoboCup3D simulation game has been more and more close to the human football game form the initial3vs3game into11vs11. Therefore, how to assign the players role and of choose formation more effectively is very important. This part we shows an arrangement of the team's formation, and assigns special duties to each player, and put forward a dynamic role assignment policy based on dynamic programming, make each player switch their roles timely according to the dynamic changes of the playing condition to meet the requirements of the game.
The above work applied to the Anhui University of RoboCup3D simulation team called DreamWing3D, and improved the team's offensive and defensive capabilities, better completed the competition.
引文
[1]Kitano, H., Tadokoro, S., Noda, I., Matsubara, H., Takahashi, T., Shinjou, A.,Shimada, S.:Robocup rescue:search and rescue in large-scale disasters as a domain for autonomous agents research[C]. In:Proc. of 1999 IEEE Int. Conf. on Systems, Man, and Cybernetics (SMC). Volume 6. (1999) 739-743 vol.6
[2]Wurman, P.R., D'Andrea, R., Mountz, M.:Coordinating hundreds of cooperative, autonomous vehicles in warehouses[C], AI Magazine 29 (2008) 9-20
[3]廉师友.人工智能技术导论[M].西安电子科技大学出版社,2002,35-41.
[4]谢涛,许建峰,张永学.仿人机器人的研究历史、现状及展望[J].机器人,2002,24(4)：367-374.
[5]鲍敦桥.仿真类人机器人设计及高层决策方法的研究[D].合肥工业大学2009.
[6]于秀丽,魏世民,廖启征.仿人机器人发展及其技术探索[J].机械工程学报,2009,45(3)：71-75.
[7]Aldebaran Robotics. NAO Robot[EB/OL].http://www.aldebaran-robotics.com.
[8]洪炳镕,阮玉峰,高庆吉等.HIT-Ⅱ型全自主足球机器人硬件系统的设计与实现[J].哈尔滨工业大学学报.2003,35(9)：1025-1028.
[9]李霞,谢涛,陈维山.基于神经网络的双足机器人逆运动学求解[J].机械设计.2003,20(4)：36-38.
[10]谢涛,徐建峰,李霞.神经网络及误差补偿在HIT-Ⅲ双足机器人步态规划中的应用.机械设计.2003,20(2)：131-133.
[11]Zhaoqing Peng, Qiang Huang et al. Online Trajectory Based on off-line Trajectory for Biped Humanoid. Proceeding of 2004 IEEE International Conference on Robotics and Biomimetics:l-5.
[12]汪光,黄强,李科杰.基于仿人机器人自身约束条件的行为调节步行控制测控与控制学报.2003,25(增刊)：45-5.
[13]Xiaojun Zhao, Qiang Huang, Zhaoqin Peng. Kinematics mapping and similarity evaluation of humanoid motion based on human motion capture. Proceedings of the IEEE/RSJ IROS'2004:840-845.
[14]A. Mackworth. On Seeing Robots[A]. In Computer Vision:Systems, Theory and Applications[C],1-13. World Scientific Press, Singapore,1992.
[15]Kitano H, Kuniyoshi Y, Noda I, et al. RoboCup:A challenge problem for AI[J]. AI Magazine,1997,18(1):73-85.
[16]石纯一,张伟,徐晋晖等.多Agent系统引论[M].电子工业出版社.2003.
[17]程显毅,王军,张俊Roboeup3D Server分析[J].计算机工程与科学,2006,28(3)：119-122
[18]The Robocup Federation. What is RoboCup[DB/OL]. http://www.Robocup.org.
[19]Patrick Riley. SPADES:System for Parallel Agent Discrete Event Simulation User's Guide and Reference Manual[EB/OL]. http://spades-sim.sourceforge.net,2003
[20]R. Smith, Open Dynamics Engine User Guide[EB/OL], http://opende.sourceforge.net, February 2006.
[21]缪克华Robocup3D足球机器人体系结构与基本技能的研究与实现[D].厦门：厦门大学控制理论与控制工程专业,2008.
[22]Saeed B. Niku机器人学导论[M].北京：电子工业出版社,2006.
[23]John J. Craig.机器人学导论[M].北京：机械工业出版社,2006.
[24]棍田秀司.仿人机器人[M].北京：清华大学出版社,2007.
[25]甘志刚,肖南峰.仿人机器人三维实时仿真系统的研究与实现[J].系统仿真学报,2007,19(11)：2444-2448,2518.
[26]洪嘉振.计算多体系统动力学[M].北京：高等教育出版社,1998：41-49.
[27]霍伟.机器人动力学与控制[M].北京：高等教育出版社,2005：23-30.
[28]刘松国,朱世强,李江波等.6R机器人实时逆运动学算法研究[J].控制理论与应用,2008,25(6)：1037-1041.
[29]周芳芳,樊晓平,赵颖.机器人逆运动学求解的可视化算法[J].计算机工程,2006,32(14)：193-195.
[30]张永学,双足机器人步态规划及步行控制研究田[D].哈尔滨：哈尔滨工业大学,2001.6.
[31]Joschka Boedecker, Klaus Dorer, et al. Simspark Users Manual [EB/OL]. http://sim-spark.sourceforge.net/wiki/index.php/Image:User-manual.pdf
[32]Kalman R. E. A new approach to linear filtering and prediction problems, transactions of the ASME [J]. Journal of Basic Engineering,1960(82):35-45.
[33]Greg Welch, Gray Bishop. An introduction to the kalman filter[R]. Chapel Hill,NC,USA,1995.
[34]Thrun S. Bayesian landmark leaning for mobile robot localization [J]. Machine Leaning,1998,33(1):41-76.
[35]方正,佟国峰,徐心和.基于贝叶斯滤波理论的自主机器人自定位方法研究[J].控制与决策,2006,(08)：841-847.
[36]刘汝佳,孙增圻.RoboCup3D仿真组中世界模型的维护[C].2005中国机器人大赛论文集.
[37]Justin Stoecker, Ubbo Visser. RoboViz:Programmable Visualization for Simulated Soccer[C]. RoboCup symposium,2011:2-14.
[38]Kalyanakrishnan, S., Stone, P.:Learning complementary multiagent behaviors:A case study. In:RoboCup 2009:Robot Soccer World Cup ⅩⅢ, Springer (2010)153-165.
[39]Stone, P., Veloso, M.:Task decomposition, dynamic role assignment, and low-bandwidth communication for real-time strategic teamwork. Artificial Intelligence 110 (1999) 241-273.
[40]Reis, L., Lau, N., Oliveira, E.:Situation based strategic positioning for coordinating a team of homogeneous agents. In Hannebauer, M., Wendler, J., Pagello, E., eds.:Balancing Reactivity and Social Deliberation in Multi-Agent Systems. Volume 2103 of Lecture Notes in Computer Science. Springer Berlin/Heidelberg (2001) 175-197.
[41]Chen, W., Chen, T.:Multi-robot dynamic role assignment based on path cost. In:2011 Chinese Control and Decision Conference (CCDC). (2011) 3721-3724.
[42]朱志强.基于动态规划与遗传模拟退火算法解定位路线安排问题[D].长安大学2006.
[43]齐驰.近似动态规划方法及其在交通中的应用[D].北京交通大学2012.
[44]MacAlpine, P., Urieli, D., Barrett, S., Kalyanakrishnan, S., Barrera, F., Lopez-Mobilia, A., S.tiurca, N., Vu, V., Stone, P.:UT Austin Villa 2011 3D SimulationTeam report. Technical Report AI11-10, The Univ. of Texas at Austin, Dept. of Computer Science, AI Laboratory (2011).
[45]MacAlpine, P., Urieli, D., Barrett, S., Kalyanakrishnan, S., Barrera,F., Lopez-Mobilia, A., S.tiurca, N., Vu, V., Stone, P.:UT Austin Villa 2011:A championagent in the RoboCup 3D soccer simulation competition. In:Proc. of 11th Int.Conf. on Autonomous Agents and Multiagent Systems (AAMAS 2012). (2012).
[2]Wurman, P.R., D'Andrea, R., Mountz, M.:Coordinating hundreds of cooperative, autonomous vehicles in warehouses[C], AI Magazine 29 (2008) 9-20
[3]廉师友.人工智能技术导论[M].西安电子科技大学出版社,2002,35-41.
[4]谢涛,许建峰,张永学.仿人机器人的研究历史、现状及展望[J].机器人,2002,24(4)：367-374.
[5]鲍敦桥.仿真类人机器人设计及高层决策方法的研究[D].合肥工业大学2009.
[6]于秀丽,魏世民,廖启征.仿人机器人发展及其技术探索[J].机械工程学报,2009,45(3)：71-75.
[7]Aldebaran Robotics. NAO Robot[EB/OL].http://www.aldebaran-robotics.com.
[8]洪炳镕,阮玉峰,高庆吉等.HIT-Ⅱ型全自主足球机器人硬件系统的设计与实现[J].哈尔滨工业大学学报.2003,35(9)：1025-1028.
[9]李霞,谢涛,陈维山.基于神经网络的双足机器人逆运动学求解[J].机械设计.2003,20(4)：36-38.
[10]谢涛,徐建峰,李霞.神经网络及误差补偿在HIT-Ⅲ双足机器人步态规划中的应用.机械设计.2003,20(2)：131-133.
[11]Zhaoqing Peng, Qiang Huang et al. Online Trajectory Based on off-line Trajectory for Biped Humanoid. Proceeding of 2004 IEEE International Conference on Robotics and Biomimetics:l-5.
[12]汪光,黄强,李科杰.基于仿人机器人自身约束条件的行为调节步行控制测控与控制学报.2003,25(增刊)：45-5.
[13]Xiaojun Zhao, Qiang Huang, Zhaoqin Peng. Kinematics mapping and similarity evaluation of humanoid motion based on human motion capture. Proceedings of the IEEE/RSJ IROS'2004:840-845.
[14]A. Mackworth. On Seeing Robots[A]. In Computer Vision:Systems, Theory and Applications[C],1-13. World Scientific Press, Singapore,1992.
[15]Kitano H, Kuniyoshi Y, Noda I, et al. RoboCup:A challenge problem for AI[J]. AI Magazine,1997,18(1):73-85.
[16]石纯一,张伟,徐晋晖等.多Agent系统引论[M].电子工业出版社.2003.
[17]程显毅,王军,张俊Roboeup3D Server分析[J].计算机工程与科学,2006,28(3)：119-122
[18]The Robocup Federation. What is RoboCup[DB/OL]. http://www.Robocup.org.
[19]Patrick Riley. SPADES:System for Parallel Agent Discrete Event Simulation User's Guide and Reference Manual[EB/OL]. http://spades-sim.sourceforge.net,2003
[20]R. Smith, Open Dynamics Engine User Guide[EB/OL], http://opende.sourceforge.net, February 2006.
[21]缪克华Robocup3D足球机器人体系结构与基本技能的研究与实现[D].厦门：厦门大学控制理论与控制工程专业,2008.
[22]Saeed B. Niku机器人学导论[M].北京：电子工业出版社,2006.
[23]John J. Craig.机器人学导论[M].北京：机械工业出版社,2006.
[24]棍田秀司.仿人机器人[M].北京：清华大学出版社,2007.
[25]甘志刚,肖南峰.仿人机器人三维实时仿真系统的研究与实现[J].系统仿真学报,2007,19(11)：2444-2448,2518.
[26]洪嘉振.计算多体系统动力学[M].北京：高等教育出版社,1998：41-49.
[27]霍伟.机器人动力学与控制[M].北京：高等教育出版社,2005：23-30.
[28]刘松国,朱世强,李江波等.6R机器人实时逆运动学算法研究[J].控制理论与应用,2008,25(6)：1037-1041.
[29]周芳芳,樊晓平,赵颖.机器人逆运动学求解的可视化算法[J].计算机工程,2006,32(14)：193-195.
[30]张永学,双足机器人步态规划及步行控制研究田[D].哈尔滨：哈尔滨工业大学,2001.6.
[31]Joschka Boedecker, Klaus Dorer, et al. Simspark Users Manual [EB/OL]. http://sim-spark.sourceforge.net/wiki/index.php/Image:User-manual.pdf
[32]Kalman R. E. A new approach to linear filtering and prediction problems, transactions of the ASME [J]. Journal of Basic Engineering,1960(82):35-45.
[33]Greg Welch, Gray Bishop. An introduction to the kalman filter[R]. Chapel Hill,NC,USA,1995.
[34]Thrun S. Bayesian landmark leaning for mobile robot localization [J]. Machine Leaning,1998,33(1):41-76.
[35]方正,佟国峰,徐心和.基于贝叶斯滤波理论的自主机器人自定位方法研究[J].控制与决策,2006,(08)：841-847.
[36]刘汝佳,孙增圻.RoboCup3D仿真组中世界模型的维护[C].2005中国机器人大赛论文集.
[37]Justin Stoecker, Ubbo Visser. RoboViz:Programmable Visualization for Simulated Soccer[C]. RoboCup symposium,2011:2-14.
[38]Kalyanakrishnan, S., Stone, P.:Learning complementary multiagent behaviors:A case study. In:RoboCup 2009:Robot Soccer World Cup ⅩⅢ, Springer (2010)153-165.
[39]Stone, P., Veloso, M.:Task decomposition, dynamic role assignment, and low-bandwidth communication for real-time strategic teamwork. Artificial Intelligence 110 (1999) 241-273.
[40]Reis, L., Lau, N., Oliveira, E.:Situation based strategic positioning for coordinating a team of homogeneous agents. In Hannebauer, M., Wendler, J., Pagello, E., eds.:Balancing Reactivity and Social Deliberation in Multi-Agent Systems. Volume 2103 of Lecture Notes in Computer Science. Springer Berlin/Heidelberg (2001) 175-197.
[41]Chen, W., Chen, T.:Multi-robot dynamic role assignment based on path cost. In:2011 Chinese Control and Decision Conference (CCDC). (2011) 3721-3724.
[42]朱志强.基于动态规划与遗传模拟退火算法解定位路线安排问题[D].长安大学2006.
[43]齐驰.近似动态规划方法及其在交通中的应用[D].北京交通大学2012.
[44]MacAlpine, P., Urieli, D., Barrett, S., Kalyanakrishnan, S., Barrera, F., Lopez-Mobilia, A., S.tiurca, N., Vu, V., Stone, P.:UT Austin Villa 2011 3D SimulationTeam report. Technical Report AI11-10, The Univ. of Texas at Austin, Dept. of Computer Science, AI Laboratory (2011).
[45]MacAlpine, P., Urieli, D., Barrett, S., Kalyanakrishnan, S., Barrera,F., Lopez-Mobilia, A., S.tiurca, N., Vu, V., Stone, P.:UT Austin Villa 2011:A championagent in the RoboCup 3D soccer simulation competition. In:Proc. of 11th Int.Conf. on Autonomous Agents and Multiagent Systems (AAMAS 2012). (2012).