摘要
以固高GRB-400型机器人和CCD组成机器人手眼系统,分析了摄像机的成像模型,采用了基于直接将图像坐标映射到机器人参考坐标的"黑箱"思想,从图像坐标直接计算出目标位置的方法,用于立体定位的摄像机手眼标定,该方法通过保持机器人末端执行器到机器人参考坐标系旋转矩阵来简化复杂的相机标定过程,最后通过实验验证了该方法的可行性,并分析了实验误差产生的原因,并提出了相应的解决措施。
This paper introduces a robotic eye-in-hand vision system which is consisted of Googol GRB-400 manipulator and CCD.Imaging model is analyzed.Based on the idea of ＂black box＂ that directly mapping the coordinates of camera to robot coordinates,a constant rotation matrix that map send-effector coordinate to robot coordinate is proposed in this paper.The constant rotation matrix simplifies the algorithm and process of object 3-D position.The feasibility of this method is verified by experimental studies.The experiment error and accuracy are analyzed,and the measure is proposed to improve the corresponding error.
出处
《计算机工程与应用》
CSCD
北大核心
2007年第33期196-199,共4页
Computer Engineering and Applications
基金
国家自然科学基金(the National Natural Science Foundation of China under Grant No.10202020)
关键词
手眼视觉
目标定位
立体视觉
eye-in-hand vision
object position
stereo vision
