摘要
基于全景视觉的定位系统采用可视人工路标作为定位依据.为避免复杂的路标制作过程,提高导航的灵活性和方便性,采用简单的彩色立方体作为定位路标.同时,为避免路标的误识别导致误定位,提出了利用路标组特征匹配的方法,提取路标组中具有尺度不变性和旋转不变性的路标组特征进行路标识别,且利用路标组里的路标信息通过3点定位原理得到探测车当前的位置,结合人机交互的路径规划判断定位的可信度,提高定位的正确度,保证车子的安全性.通过软件仿真和现场实验,证明了该视觉定位系统可避免以往基于可视路标定位存在的不稳定性,大幅度减少了定位错误概率.
In order to realize explorers＇ visual navigation mission, localization is required. The omni-directional vision system uses artificial landmarks to realize self-localization. Simple color cubes are used as artificial landmarks to avoid complex landmarks making. The method of matching based on group features is used to identify different landmarks because a group of landmarks has the features with scale invariability and rotation invariability. The position of the explorer is obtained according to groups of landmarks using the theory of three-point localization. The planned path is used to judge whether the localization is right. The localization system uses simple landmarks with unique rules and conditions to realize self-localization, avoiding the instability of the previous localization based on landmarks, decreasing the mistakes of localization.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2009年第6期871-874,共4页
Journal of Shanghai Jiaotong University
基金
教育部重大项目培育资金项目(708035)
关键词
全景视觉
定位系统
路标识别
定位算法
omni-directional vision
localization system
landmarks recognitions
location algorithm
