摘要
为了解决中型组比赛环境下的足球机器人自定位问题,本文提出一种新的基于全向视觉的自定位方法.该定位方法首先从全景图像中提取出场上白线的对应点,并将其作为机器人的观测信息,然后计算观测信息与静态地图的匹配误差,且根据匹配误差的大小决定是否重新选取主位姿,最后利用梯度优化算法修正主位姿,并以主位姿信息作为定位结果.仿真结果说明了该定位方法的有效性.
To solve the problem of robot self-localization in the Robocop Middle Size league, a new method for robot self-localization based on omni-direetional vision sensor is proposed. First, points on field lines as visual data are extracted from the image. Then, the error of matching between visual data and map is calculated to decide whether choose main pose again. Finally, the main pose is corrected by use of gradient descent algorithm. Simulation results demonstrate that the effectiveness of the proposed method.
出处
《微计算机信息》
2009年第8期270-272,共3页
Control & Automation
关键词
自定位
地图匹配
梯度优化
serf-localization
map-matching
gradient optimization
