足球仿人机器人的视觉系统的设计与研究
|
摘要
足球仿人机器人比赛中的视觉系统是机器人获取信息的主要来源,能否准确快速的识别场上的目标物体是机器人设计的基石,因此仿人机器人的视觉系统设计过程中的目标识别和目标跟踪有着及其重要的意义。本文针对足球仿人机器人比赛中的具体问题,如单目测距系统的构建、如过多碰撞问题主要做了以下的工作:
首先,在基于HSI颜色空间的多目标识别的基础上,结合正运动学运动原理和单孔成像原理构建了足球仿人机器人的单目测距系统,并与RGB颜色空间上的识别效果进行了比较。并用实验分析了单目测距系统影响测距效果的各种因素,如俯仰角、机器人的动态身高、目标点的选取等。
其次,针对足球仿人机器人比赛中碰撞过多的情况,介绍了人工势场法在机器人足球路径规划中的应用,针对人工势场法容易陷入局部最小的缺陷,通过增加一个随机角度生成来改进人工势场法。另外提出一种基于视觉信息的仿人足球机器人的局部路径规划方法,对机器人获取的图像进行基于颜色阈值的扫描线分割算法来获取表征障碍物的最远点,利用单目摄像机成像原理来获取最远点在机器人坐标系中的位置。为解决局部路径规划中遇到的避障问题,对机器人视觉范围进行扇区的划分,通过扇区选择函数得到可行扇区的序号来选择当前的路径方向。仿真实验证明该方法的可行性以及满足机器人足球比赛的实时性,同时分别在平均消耗时间和碰撞次数上对改进的人工势场法和基于扇区模型的方法进行了比较。
最后,介绍了HfutEngine MSRS球队的视觉框架的设计以及上述方法在球队中的应用。
Vision system is main method for humanoid soccer robot catching information in humanoid soccer game. The footstone of robot designing is lying to whether the vision system can recognize and track goal accurately and fast. So the design and research of vision system is significant. The thesis aim to detail problems such as monocular distance measurement and so on, main works are as follows:
First, the thesis build monocular ranging based on HSI color space besides forward kinematics and pinhole principle, and compares the result of reorganization between RGB and HSI color space. The experiment analyzes different elements which influence ranging such as angle of pitch and height of monocular, chosen of feature point of object when robots move.
Second, Simulated humanoid soccer has many collisions, the thesis introduce artificial potential field method for local path planning and add random angle method to avoid local minimum. Besides, the thesis present a local path planning of humanoid soccer robot based on monocular vision. It gets the far points which represent the obstacles by segmenting the image based on color threshold scan-line method in HSI color space. The value of robot coordination of far point is calculated by monocular vision imaging principle. To deal with the obstacle avoidance, the field of view of robot is divided to several sectors, the current direction of robot’s is determined by the number of selected sectors .The simulated experiment prove its feasibility and suit for real-time of robot soccer match, compare with artificial potential field(APF) in time-consuming and times of success.
Last, the thesis introduces the implementation and application of HfutEngine MSRS humanoid soccer team.
首先,在基于HSI颜色空间的多目标识别的基础上,结合正运动学运动原理和单孔成像原理构建了足球仿人机器人的单目测距系统,并与RGB颜色空间上的识别效果进行了比较。并用实验分析了单目测距系统影响测距效果的各种因素,如俯仰角、机器人的动态身高、目标点的选取等。
其次,针对足球仿人机器人比赛中碰撞过多的情况,介绍了人工势场法在机器人足球路径规划中的应用,针对人工势场法容易陷入局部最小的缺陷,通过增加一个随机角度生成来改进人工势场法。另外提出一种基于视觉信息的仿人足球机器人的局部路径规划方法,对机器人获取的图像进行基于颜色阈值的扫描线分割算法来获取表征障碍物的最远点,利用单目摄像机成像原理来获取最远点在机器人坐标系中的位置。为解决局部路径规划中遇到的避障问题,对机器人视觉范围进行扇区的划分,通过扇区选择函数得到可行扇区的序号来选择当前的路径方向。仿真实验证明该方法的可行性以及满足机器人足球比赛的实时性,同时分别在平均消耗时间和碰撞次数上对改进的人工势场法和基于扇区模型的方法进行了比较。
最后,介绍了HfutEngine MSRS球队的视觉框架的设计以及上述方法在球队中的应用。
Vision system is main method for humanoid soccer robot catching information in humanoid soccer game. The footstone of robot designing is lying to whether the vision system can recognize and track goal accurately and fast. So the design and research of vision system is significant. The thesis aim to detail problems such as monocular distance measurement and so on, main works are as follows:
First, the thesis build monocular ranging based on HSI color space besides forward kinematics and pinhole principle, and compares the result of reorganization between RGB and HSI color space. The experiment analyzes different elements which influence ranging such as angle of pitch and height of monocular, chosen of feature point of object when robots move.
Second, Simulated humanoid soccer has many collisions, the thesis introduce artificial potential field method for local path planning and add random angle method to avoid local minimum. Besides, the thesis present a local path planning of humanoid soccer robot based on monocular vision. It gets the far points which represent the obstacles by segmenting the image based on color threshold scan-line method in HSI color space. The value of robot coordination of far point is calculated by monocular vision imaging principle. To deal with the obstacle avoidance, the field of view of robot is divided to several sectors, the current direction of robot’s is determined by the number of selected sectors .The simulated experiment prove its feasibility and suit for real-time of robot soccer match, compare with artificial potential field(APF) in time-consuming and times of success.
Last, the thesis introduces the implementation and application of HfutEngine MSRS humanoid soccer team.
引文
[1] Karel ?apek. http://zh.wikipedia.org/wiki/卡雷尔·恰佩克.
[2]杨莹,丁堃，许侃.国际机器人科学知识前沿演化的可视化分析[C].科学理论与科学计量学探索—全国科学技术学暨科学理论与学科建设2008年联合年会论文集.中国安徽黄山,2008:168-173.
[3] http://baike.baidu.com/view/630520.htm.
[4] http://baike.baidu.com/view/2307509.html
[5]初苗苗. RoboCup中型组足球机器人硬件平台设计与实现及视觉系统若干问题研究[D].硕士学位论文.山东大学,2009.
[6]欧宗瑛袁野张艳珍.基于颜色信息足球机器人视觉跟踪算法.大连理工大学学报.2000.40(6).p729-732
[7]Jungel,Matthias.Lecture Notes in Artificial Intelligence.Robot Soccer World Cup VIII.Lisbon Portugal.2004.Springer Verlag.2005.p209-220
[8]李会猛,陈家乾,诸静.基于自适应颜色识别技术的快速多目标跟踪方法.计算机工程与应用.2005.30. p59-62
[9]张艳珍,袁野,巩玉国等.基于区域投影的微型足球机器人系统识别.机械科学与技术.2001.20(2).P318-320
[10]曹丹华,邹伟,吴裕斌.基于背景图像差分的运动人体检测[J].光电工程,2007,34(6):107—111.
[11]王栓,艾海舟,何克忠.基于差分图像的多运动目标的检测与跟踪[J].中国图像图形学报,1999,4(6):470—475.
[12] Zhang Z.A flexible new technique for camera calibration[J].—IEE—E Transaction on Pattem Analysis and Machine Intelligence,2000,22(11):1330—1334.
[13] Peng Xiaoyan,Zhang Yong,Tan Hefiang.Improve on Recognition Algorithms of RoboCup[C].ShangHai:Proceedings of the WCICA’02,2002:515—518.
[14] Latombe J C. Robot motion planning [M]. Boston, Kluwer Academic Publishers, 1991.
[15] Khatib O. Real2time obstacle avoidance for manipula2tors and mobile robot [J]. The International Journal of Robotic Research, 1986, 5(1):90 - 98.
[16] Park M G,Jeon J H ,Lee M C ,Obstacle avoidance formobile robots using artificial potential field approach with simulated annealing[A] . ISIE[C]. Pusan, Korea, 2001:1530 - 1535.
[17] Vadakkepat P ,Tan K C ,Wang M L. Devolutionary artificial potential fields and their application in real time robot path planning[A]. Congress on EvolutionaryComputation[C]. 2000:256 - 263.
[18]谢凤英,赵主培.Visual C++数字图像处理[M].北京:电子工业出版社,2008.
[19]夏良正.数字图像处理(修订版)[M].南京:东南大学出版社,2001.
[20]Rafael C.Gonzalez,Richard E.Woods著;阮秋琦,阮宇智等译.数字图像处理(第二版)[M].北京:电子工业出版社,2003.
[21] http://www.hzzhibang.com.cn/hzzhibang_Article_16525.html.
[22] ALBIOL, A., TORRES, L., AND DELP, E. J. 2001. Optimum color spaces for skin detection. In Proceedings of the International Conference on Image Processing, vol. 1, 122–124.
[23]章毓晋.图象分割.北京:科学出版社,2001.1~5
[24]张讲社,梁怡,徐宗本.基于视觉系统的聚类算法.计算机学报,2001,24(5):496~451
[25]Kehtarnavaz N., Monaco J.Nimtschek J.et al.Color image segmentation using multi-scale clustering. Image Analysis and Interpretation, 1998 IEEE Southwest Symposium on, 1998:142~147
[26] Jain, S. and R. M. Neal (2004). A split-merge Markov chain Monte Carlo procedure for the Dirichlet Process mixture model. Journal of Computational and Graphical Statistics 13(1), in press.
[27] G. J. Klinker, S. A. Shafer, and T. Kanade, A Physical approach to color image understanding, Int. J. Comput. Vision 4(1), 1990, 7–38.
[28] Lindeberg, T. (1994). Scale-space theory in computer vision. Dordrecht:Kluwer Academic.
[29] Vieira Lopes N, Mogadouro do Couto PA, Bustince H, and Melo-Pinto P. Automatic histogram threshold using fuzzy measures. IEEE Trans Image Process. 2010 Jan; 19(1):199-204.
[30]R.Ohlander, K.Price, D.R.Reddy. Picture segmentation using cursive region splitting method. Computer Graphics and Image Processing, 1980, vol.13:222~224
[31]Liu L., Sclaroff S.Region segmentation via deformable model-guided split andmerge, Computer Vision, 2001.ICCV 2001.Proceedings.Eighth IEEE International Conference on, 2001, 1(7-14):98~104
[32] J. van de Weijer, C. Schmid, J. Verbeek, Using high-level visual information for color constancy, in: Proceedings of the IEEE 14th International Conference on Computer Vision, 2007, pp. 1–8.
[33]谢云,杨宜民.自主足球机器人的单目视觉自定位方法[J].微电子学与计算机.2005, 22(10): 129-132.
[34]梶田秀司.仿人机器人[M].管贻生,译.北京:清华大学出版社,2007.
[35]周贤,等.基于图像边缘能量的自动聚焦算法[J ] .光学技术,2006 , 32 (2) :213—218.
[36]刘华军,杨静宇,陆建峰等.移动机器人运动规划研究综述[J].中国工程科学,2006,8(1):85-94.
[37]O.Khatib.Real-time Obstacle Avoidance for Manipulator and Mobile Robots. The International Journal of Robotics Research, 1986, 5(1):90-98.
[38]周郭许,唐西林.基于栅格模型的机器人路径规划快速算法[J].计算机工程与应用,2006,42(21):196-199.
[39]Hairong Xiao,Li Liao Fengyu Zhou.Mobile Robot Path Planning Based on Q-ANN.Automation and Logistics. IEEE International Conference.Jina, China 2007, pp2650~2654.
[40]Cheng-Dong Wu,Ying Zhang,Meng-Xin Li,Yong Yue.A Rough Set GA-based Hybrid Method for Robot Path Planning[J].International Jounrnal of Automation and Computing,2006,3(1):29-34.
[41] I F Akyildiz, E Ekici, M D Bender. MLSR: a novel routing algorithm for multilayered satellite IP networks [J]. IEEE/ACM Transactions on Networking (S1063-6692), 2002, 10(3): 411-424.
[42]李晓丽,谢敬,傅卫平,等.一种改进势场法在多移动机器人避碰规划中的应用[J].计算机工程与应用, 2005, 41( 17) : 56- 58.
[43]张培艳,吕恬生.基于模拟退火-人工势场法的足球机器人路径规划研究[J].机械科学与技术, 2003, 22( 4) : 547- 549.
[44] [6] Borenstein J, Koren Y.Real- time obstacle avoidance for fast mobile robots in cluttered environments[C]//the 1990 IEEE International Conference on Robotics and Automation. Cincinnati Ohio: IEEE, 1990: 572- 577.
[45]黄炳强、曹广益,基于人工势场法的移动机器人路径规划研究[J].计算机工程与应用,2006,No.27:26-28.
[46] http://www.msuniversity.edu.cn/m_RepositoryIntro/Detail.aspx?id=744
[2]杨莹,丁堃，许侃.国际机器人科学知识前沿演化的可视化分析[C].科学理论与科学计量学探索—全国科学技术学暨科学理论与学科建设2008年联合年会论文集.中国安徽黄山,2008:168-173.
[3] http://baike.baidu.com/view/630520.htm.
[4] http://baike.baidu.com/view/2307509.html
[5]初苗苗. RoboCup中型组足球机器人硬件平台设计与实现及视觉系统若干问题研究[D].硕士学位论文.山东大学,2009.
[6]欧宗瑛袁野张艳珍.基于颜色信息足球机器人视觉跟踪算法.大连理工大学学报.2000.40(6).p729-732
[7]Jungel,Matthias.Lecture Notes in Artificial Intelligence.Robot Soccer World Cup VIII.Lisbon Portugal.2004.Springer Verlag.2005.p209-220
[8]李会猛,陈家乾,诸静.基于自适应颜色识别技术的快速多目标跟踪方法.计算机工程与应用.2005.30. p59-62
[9]张艳珍,袁野,巩玉国等.基于区域投影的微型足球机器人系统识别.机械科学与技术.2001.20(2).P318-320
[10]曹丹华,邹伟,吴裕斌.基于背景图像差分的运动人体检测[J].光电工程,2007,34(6):107—111.
[11]王栓,艾海舟,何克忠.基于差分图像的多运动目标的检测与跟踪[J].中国图像图形学报,1999,4(6):470—475.
[12] Zhang Z.A flexible new technique for camera calibration[J].—IEE—E Transaction on Pattem Analysis and Machine Intelligence,2000,22(11):1330—1334.
[13] Peng Xiaoyan,Zhang Yong,Tan Hefiang.Improve on Recognition Algorithms of RoboCup[C].ShangHai:Proceedings of the WCICA’02,2002:515—518.
[14] Latombe J C. Robot motion planning [M]. Boston, Kluwer Academic Publishers, 1991.
[15] Khatib O. Real2time obstacle avoidance for manipula2tors and mobile robot [J]. The International Journal of Robotic Research, 1986, 5(1):90 - 98.
[16] Park M G,Jeon J H ,Lee M C ,Obstacle avoidance formobile robots using artificial potential field approach with simulated annealing[A] . ISIE[C]. Pusan, Korea, 2001:1530 - 1535.
[17] Vadakkepat P ,Tan K C ,Wang M L. Devolutionary artificial potential fields and their application in real time robot path planning[A]. Congress on EvolutionaryComputation[C]. 2000:256 - 263.
[18]谢凤英,赵主培.Visual C++数字图像处理[M].北京:电子工业出版社,2008.
[19]夏良正.数字图像处理(修订版)[M].南京:东南大学出版社,2001.
[20]Rafael C.Gonzalez,Richard E.Woods著;阮秋琦,阮宇智等译.数字图像处理(第二版)[M].北京:电子工业出版社,2003.
[21] http://www.hzzhibang.com.cn/hzzhibang_Article_16525.html.
[22] ALBIOL, A., TORRES, L., AND DELP, E. J. 2001. Optimum color spaces for skin detection. In Proceedings of the International Conference on Image Processing, vol. 1, 122–124.
[23]章毓晋.图象分割.北京:科学出版社,2001.1~5
[24]张讲社,梁怡,徐宗本.基于视觉系统的聚类算法.计算机学报,2001,24(5):496~451
[25]Kehtarnavaz N., Monaco J.Nimtschek J.et al.Color image segmentation using multi-scale clustering. Image Analysis and Interpretation, 1998 IEEE Southwest Symposium on, 1998:142~147
[26] Jain, S. and R. M. Neal (2004). A split-merge Markov chain Monte Carlo procedure for the Dirichlet Process mixture model. Journal of Computational and Graphical Statistics 13(1), in press.
[27] G. J. Klinker, S. A. Shafer, and T. Kanade, A Physical approach to color image understanding, Int. J. Comput. Vision 4(1), 1990, 7–38.
[28] Lindeberg, T. (1994). Scale-space theory in computer vision. Dordrecht:Kluwer Academic.
[29] Vieira Lopes N, Mogadouro do Couto PA, Bustince H, and Melo-Pinto P. Automatic histogram threshold using fuzzy measures. IEEE Trans Image Process. 2010 Jan; 19(1):199-204.
[30]R.Ohlander, K.Price, D.R.Reddy. Picture segmentation using cursive region splitting method. Computer Graphics and Image Processing, 1980, vol.13:222~224
[31]Liu L., Sclaroff S.Region segmentation via deformable model-guided split andmerge, Computer Vision, 2001.ICCV 2001.Proceedings.Eighth IEEE International Conference on, 2001, 1(7-14):98~104
[32] J. van de Weijer, C. Schmid, J. Verbeek, Using high-level visual information for color constancy, in: Proceedings of the IEEE 14th International Conference on Computer Vision, 2007, pp. 1–8.
[33]谢云,杨宜民.自主足球机器人的单目视觉自定位方法[J].微电子学与计算机.2005, 22(10): 129-132.
[34]梶田秀司.仿人机器人[M].管贻生,译.北京:清华大学出版社,2007.
[35]周贤,等.基于图像边缘能量的自动聚焦算法[J ] .光学技术,2006 , 32 (2) :213—218.
[36]刘华军,杨静宇,陆建峰等.移动机器人运动规划研究综述[J].中国工程科学,2006,8(1):85-94.
[37]O.Khatib.Real-time Obstacle Avoidance for Manipulator and Mobile Robots. The International Journal of Robotics Research, 1986, 5(1):90-98.
[38]周郭许,唐西林.基于栅格模型的机器人路径规划快速算法[J].计算机工程与应用,2006,42(21):196-199.
[39]Hairong Xiao,Li Liao Fengyu Zhou.Mobile Robot Path Planning Based on Q-ANN.Automation and Logistics. IEEE International Conference.Jina, China 2007, pp2650~2654.
[40]Cheng-Dong Wu,Ying Zhang,Meng-Xin Li,Yong Yue.A Rough Set GA-based Hybrid Method for Robot Path Planning[J].International Jounrnal of Automation and Computing,2006,3(1):29-34.
[41] I F Akyildiz, E Ekici, M D Bender. MLSR: a novel routing algorithm for multilayered satellite IP networks [J]. IEEE/ACM Transactions on Networking (S1063-6692), 2002, 10(3): 411-424.
[42]李晓丽,谢敬,傅卫平,等.一种改进势场法在多移动机器人避碰规划中的应用[J].计算机工程与应用, 2005, 41( 17) : 56- 58.
[43]张培艳,吕恬生.基于模拟退火-人工势场法的足球机器人路径规划研究[J].机械科学与技术, 2003, 22( 4) : 547- 549.
[44] [6] Borenstein J, Koren Y.Real- time obstacle avoidance for fast mobile robots in cluttered environments[C]//the 1990 IEEE International Conference on Robotics and Automation. Cincinnati Ohio: IEEE, 1990: 572- 577.
[45]黄炳强、曹广益,基于人工势场法的移动机器人路径规划研究[J].计算机工程与应用,2006,No.27:26-28.
[46] http://www.msuniversity.edu.cn/m_RepositoryIntro/Detail.aspx?id=744