全维视觉机器人距离标定方法*

作者：高伟敬,韩光胜 日期：2008-02-25/span> 浏览：3952 查看PDF文档

全维视觉机器人距离标定方法*

高伟敬,韩光胜
（北京工业大学 电子信息与控制工程学院，北京 100022）

摘要：通过分析全维视觉系统的特点，总结出了常用全维视觉系统距离的标定方法，指出了各自的应用范围及优缺点。并针对应用范围最广的插值标定方法存在“单个采样点对标定精度影响较大”的缺点，提出了一种利用人工神经网络进行标定的新方法。实验证明，该距离标定方法不但可以节省标定时间，还可以提高标定精度。
关键词：全维视觉；距离标定；BP神经网络
中图分类号：TP391文献标识码：A文章编号：1001-4551(2008)01-0037-04

Methods of distance calibration for omnidirectional robot
GAO Weijing, HAN Guangsheng
(College of Electronic Information & Control Engineering, Beijing University of Technology, Beijing 100022, China)
Abstract: Several widely used calibration methods as well as their applied range, advantage and disadvantage were summarized based on analyzing the characteristic of the omnidirectional vision system. Because of the disadvantage of the embolic method, which was the single sample could badly affect the calibrating precision, a new calibrating method was proposed, which used the neural network to make distance calibration. The experiment shows that the method can save the calibration time and improve the calibration precision.
Key words: omnidirectional vision; distance calibration; BP neural network
参考文献(Reference):
［1］贾云得.机器视觉［M］.北京:科学出版社,2001.
［2］刘伟，刘斐，郑志强.用于机器人足球赛的全景视觉设计仿真［J］.计算机仿真，2005，22（11）：190－192.
［3］PALETTA L, FRINTROP S, HERTZBERZ J. Robust Localization using Context in Omnidirectional Imaging［C］. International conference on Robotics and Automation,2001.
［4］LSHIGURO H. Development of Lowcost Compact Omnidirectional Vision Sensors and their Applications［C］//Proc of Intl Conf on Information Systems, Analysis, and Synthesis,1998,9(2):433-439.
［5］易大义,沈云宝，李有法.计算方法：第2版［M］.杭州:浙江大学出版社,2002.
［6］朱大奇.人工神经网络原理及应用［M］.北京:科学出版社,2006.