一种新的双目立体视觉伺服控制模型及实验研究  

A new model of binocular vision in image-based visual servoing robot control

作　　者：李惠光[1,2] 康振波[1,2] 冷春辉[1,2] 金梅[1,2] 邹立颖[1,2] 

机构地区：[1]燕山大学电气工程学院,河北秦皇岛066004 [2]燕山大学工业计算机控制工程河北省重点实验室,河北秦皇岛066004

出　　处：《燕山大学学报》2011年第3期215-222,共8页Journal of Yanshan University

基　　金：河北省自然科学基金资助项目(F2008000860)

摘　　要：针对目前手眼视觉伺服系统模型中普遍存在深度信息估计的问题,提出了一种用于双目视觉伺服控制的模型,该模型避免了深度值的测量与估计,提高了系统的控制性能,从而解决了未知物点的深度信息估计的问题。本文应用此模型,单独考虑机器人的运动学特性,设计了机器人末端执行器进行定位控制的控制器,仿真结果验证了该模型与控制算法的有效性;进一步提出了模型使用的改进算法,使该模型更具有宽泛的实用性;并且,进行了基于MOTOMAN UP6型机器人的双目视觉伺服控制实验,实验结果验证了该模型在实际控制工程中的有效性、可行性。In most eye-in-hand visual servoing systems,the depth information must be given,because the traditional image Jacobian matrix always needs one.In this paper,a new binocular stereo visual servo model that the depth information can be ignored is pro-posed.In the new model,three image coordinate variables are introduced and which counteract the depth coordinate variable,so it is not needed to measure or estimate the depth information about the target.The depth estimation problem is completely solved and the control performance of the servo system is improved in the new model.By using this model and considering the kinematics characteristics of the robot separately,a positioning controller of a robotic end effector has been designed.The effectiveness of the controller has been verified via simulation and put forward the improvement algorithm with wider practicability.Furthermore,the binocular vision control experiment has been done by means of the MOTOMAN UP6 robotic manipulator.The experimental result shows that the new model is effective and feasible in practical control project.

关 键 词：手眼系统 深度信息 双目视觉伺服 机器人运动学 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]