基于全景视觉的多机器人协调定位
发布时间：2019-01-02 09:18
【摘要】：随着全球科技的不断进步,移动机器人技术得到了快速的发展。目前移动机器人已经逐渐应用于各行各业,并且发挥着积极重要的作用。但随着移动机器人应用环境复杂性的不断提高,使得以往单机器人的执行能力不能满足复杂任务的要求,所以多机器人合作技术成为当今机器人研究的热点之一。多机器人之间的合作离不开机器人之间位置信息的获取,因此本论文主要研究基于全景视觉的多机器人协调定位系统,该系统旨在完成多机器人之间识别与定位以及提高多机器人之间的定位精度。本文通过实际搭建的移动机器人平台验证了机器人目标识别与定位、自适应令牌环网络通信、数据融合定位等相关方法的可靠性和稳定性。具体的研究内容如下:首先,根据移动机器人运动灵活、价格便宜、易于控制等要求,本文设计并搭建了移动机器人运动平台,在该运动平台的基础上,对机器人整体方案进行具体设计,其中包括:机器人整体软硬件框架设计、传感器选型与配置、底层运动控制系统设计、自适应令牌环网通信设计。其次,关于机器人全景视觉系统,本文给出了视觉处理的具体过程。在目标体的识别过程中,本文设计了一种易识别、结构简单、可替换的机器人目标体,同时给出了一种基于极坐标下的全景成像识别方法,该方法简单有效,可以快速识别多个目标机器人的特征点。针对识别出的机器人目标体的特征点,本文提出了一种改进的K-means算法,该算法可以快速有效地得出其它机器人的坐标和角度。通过实验标定,可得出其它机器人相对自身的实际距离和角度,同时还验证了令牌环网络的实时性和稳定性。再次,对于多机器人协调定位以及定位精度提高方面,本文建立了基于航位推算的机器人底层运动模型。结合机器人之间的网络信息共享以及视觉系统下多目标的情况,本文利用电子罗盘坐标统一性的特点对多机器人目标进行解算匹配,从而建立了机器人观测模型,结合机器人的运动模型和观测模型,本文建立了多机器人之间定位状态方程,最后利用扩展卡尔曼(EKF)算法对各个模型数据进行融合,通过实验验证了该方法提高了机器人之间的定位精度。最后,本文对多机器人协作定位系统进行了总结与展望,希望通过本文提出的方案与算法,为后期研究多机器人合作奠定基础。
[Abstract]:With the continuous progress of global science and technology, mobile robot technology has been rapid development. At present, mobile robots have been gradually used in various industries, and play an active and important role. However, with the increasing complexity of mobile robot application environment, the execution ability of single robot can not meet the requirements of complex tasks in the past, so multi-robot cooperation technology has become one of the hotspots of robot research. The cooperation between multi-robots can not be separated from the acquisition of position information between robots. Therefore, this paper mainly studies the coordinated localization system of multi-robots based on panoramic vision. The purpose of this system is to realize the identification and localization of multi-robots and to improve the positioning accuracy of multi-robots. In this paper, the reliability and stability of robot target recognition and localization, adaptive token ring network communication and data fusion localization are verified by the mobile robot platform. The specific research contents are as follows: firstly, according to the requirements of mobile robot, such as flexible motion, cheap price and easy control, this paper designs and builds the mobile robot motion platform, based on the motion platform, The overall scheme of the robot is designed, including: the overall hardware and software framework design, sensor selection and configuration, the bottom motion control system design, adaptive token ring network communication design. Secondly, regarding the robot panoramic vision system, this paper gives the concrete process of vision processing. In the process of object recognition, this paper designs a robot object which is easy to recognize, simple in structure and replaceable. At the same time, a panoramic image recognition method based on polar coordinates is presented, which is simple and effective. The feature points of multiple target robots can be quickly identified. In this paper, an improved K-means algorithm is proposed for identifying the characteristic points of the target body of the robot. This algorithm can quickly and effectively obtain the coordinates and angles of other robots. Through the experimental calibration, the actual distance and angle of other robots relative to their own can be obtained, and the real-time and stability of the token ring network can also be verified. Thirdly, for the coordination of multi-robot positioning and the improvement of positioning accuracy, this paper establishes the robot bottom motion model based on dead-reckoning. Combined with the sharing of network information among robots and the situation of multi-targets under visual system, this paper makes use of the unity of electronic compass coordinates to solve and match the multi-robot targets, and then establishes the robot observation model. Combined with the motion model and observation model of the robot, the state equation of multi-robot localization is established in this paper. Finally, the extended Kalman (EKF) algorithm is used to fuse the data of each model. The experimental results show that this method can improve the positioning accuracy of robots. Finally, this paper summarizes and prospects the multi-robot cooperative positioning system, hoping to lay a foundation for the later study of multi-robot cooperation through the proposed scheme and algorithm.
【学位授予单位】：安徽工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242
[Abstract]:With the continuous progress of global science and technology, mobile robot technology has been rapid development. At present, mobile robots have been gradually used in various industries, and play an active and important role. However, with the increasing complexity of mobile robot application environment, the execution ability of single robot can not meet the requirements of complex tasks in the past, so multi-robot cooperation technology has become one of the hotspots of robot research. The cooperation between multi-robots can not be separated from the acquisition of position information between robots. Therefore, this paper mainly studies the coordinated localization system of multi-robots based on panoramic vision. The purpose of this system is to realize the identification and localization of multi-robots and to improve the positioning accuracy of multi-robots. In this paper, the reliability and stability of robot target recognition and localization, adaptive token ring network communication and data fusion localization are verified by the mobile robot platform. The specific research contents are as follows: firstly, according to the requirements of mobile robot, such as flexible motion, cheap price and easy control, this paper designs and builds the mobile robot motion platform, based on the motion platform, The overall scheme of the robot is designed, including: the overall hardware and software framework design, sensor selection and configuration, the bottom motion control system design, adaptive token ring network communication design. Secondly, regarding the robot panoramic vision system, this paper gives the concrete process of vision processing. In the process of object recognition, this paper designs a robot object which is easy to recognize, simple in structure and replaceable. At the same time, a panoramic image recognition method based on polar coordinates is presented, which is simple and effective. The feature points of multiple target robots can be quickly identified. In this paper, an improved K-means algorithm is proposed for identifying the characteristic points of the target body of the robot. This algorithm can quickly and effectively obtain the coordinates and angles of other robots. Through the experimental calibration, the actual distance and angle of other robots relative to their own can be obtained, and the real-time and stability of the token ring network can also be verified. Thirdly, for the coordination of multi-robot positioning and the improvement of positioning accuracy, this paper establishes the robot bottom motion model based on dead-reckoning. Combined with the sharing of network information among robots and the situation of multi-targets under visual system, this paper makes use of the unity of electronic compass coordinates to solve and match the multi-robot targets, and then establishes the robot observation model. Combined with the motion model and observation model of the robot, the state equation of multi-robot localization is established in this paper. Finally, the extended Kalman (EKF) algorithm is used to fuse the data of each model. The experimental results show that this method can improve the positioning accuracy of robots. Finally, this paper summarizes and prospects the multi-robot cooperative positioning system, hoping to lay a foundation for the later study of multi-robot cooperation through the proposed scheme and algorithm.
【学位授予单位】：安徽工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242
【参考文献】
相关期刊论文 前10条
1 侯伟龙;;基于模糊PID的助行机器人调速系统的研究[J];计算机测量与控制;2017年01期
2 徐明钊;于海涛;杨春;翟章明;王旭刚;;基于陀螺仪和码盘的自主定位机器人系统[J];兵工自动化;2016年10期
3 刘辛军;于靖军;王国彪;赖一楠;何柏岩;;机器人研究进展与科学挑战[J];中国科学基金;2016年05期
4 吴晓军;郭亚雪;;一种圆柱物体360°全景成像装置设计方法[J];光学学报;2015年09期
5 蔡磊涵;;CMOS图像传感器在监控市场主导地位提升[J];中国安防;2015年10期
6 郝晨;李航;;扩展卡尔曼和无迹卡尔曼滤波应用对比研究[J];沈阳师范大学学报(自然科学版);2015年02期
7 张铁;陈伟华;邹焱飚;马琼雄;;移动机器人里程计系统误差的校正方法[J];华南理工大学学报(自然科学版);2014年09期
8 昝云龙;薛颂东;曾建潮;;绝对定位机制下目标位置估计辅助的群机器人搜索[J];模式识别与人工智能;2013年04期
9 张学习;杨宜民;;基于多传感器信息融合的移动机器人快速精确自定位[J];控制理论与应用;2011年03期
10 黄文力;马礼;王玉超;;嵌入式Internet中TCP/IP协议栈的研究与设计[J];电脑开发与应用;2008年01期
相关博士学位论文 前3条
1 王勇;移动机器人的定位能力估计及定位[D];上海交通大学;2014年
2 蔡云飞;多机器人协作定位及系统架构研究[D];南京理工大学;2011年
3 席志红;全方位视觉技术及其在智能移动机器人等领域的应用研究[D];哈尔滨工程大学;2006年
相关硕士学位论文 前7条
1 胡胜豪;室外移动机器人的定位与运动控制研究[D];哈尔滨工业大学;2016年
2 闵光星;基于人类视觉系统的视频质量评价模型研究[D];吉林大学;2012年
3 汤国望;交互式移动机器人系统的设计与实现[D];华中科技大学;2012年
4 李江辉;基于Linux平台的Ad hoc网络应用与设计实现[D];武汉科技大学;2011年
5 陈U，
本文编号：2398314
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2398314.html
