| With the improvement of the automation level of industrial robots,the automatic processing participated by industrial robots will become the leading part of the future industrial development.Grasping and handling the goods by visual guidance robot is of great significance to improve the automation level of production line and broaden the application range of robot.Thus,the concept of"bin-picking"was introduced,which means the robot completed the grasping of scattered objects based on the assistance of3D vision system.One of the key technologies for the bin-picking system is the position relation calibration between the industrial robot and the vision sensor.This paper studied the 3d hand-eye calibration technology for bin-picking system.Specific work includes the following aspects:(1)based on the basic situation of grasping objects as loose rods,the eye-to-hand hand-eye model was determined,and the hand-eye scheme based on the line-structured light robot hand-eye system was designed.This paper introduced a hand Eye calibration model of eye-to-hand line structured light system based on theP_R~i(28)XP_C~i model,the hand eye relation cloud be obtained,as long as the coordinate information of the target characteristic points under the coordinate system of the online structured light vision sensor and the base coordinate system of the robot cloud be obtained.(2)In the process of obtaining the coordinate information of the spherical center of the standard ball target under the coordinate system of the online structured light vision sensor,a standard spherical point cloud spherical center extraction method based on the consistency of random sampling was proposed,which solves the problem of big error caused by not considering outliers in the fitting method of least square method.(3)In the process of obtaining the coordinate information of the center of the standard ball target in the robot base coordinate system,a non-contact method based on monocular vision was proposed to calibrate the position relationship between the center of the robot flange and the center of the standard ball target.This method is more accurate and reliable than simply relying on fixed mechanical dimensions.(4)An improved ICP algorithm based on the"centripetal vector"of point set was proposed to solve the hand-eye matrix.Compared with the traditional ICP algorithm,the improved ICP algorithm in this paper used centripetal vector as the index in the nearby point pairing and the nearby point preprocessing of the point set,which improved the accuracy of the nearby point pairing,and improved the quality of the initial position of the point set,and provided a good foundation for the subsequent solving of the hand-eye matrix.(5)A hand eye calibration experiment platform based on a structured light vision sensor and a robot was built,a hand-eye calibration software was written based on the improved ICP algorithm.The traditional ICP algorithm and the improved ICP algorithm were respectively used to solve the hand eye matrix,and the 3d positioning accuracy experiment was designed to verify the accuracy of the hand eye matrix.The experimental results show that the iteration times and iteration time of the improved ICP algorithm are further shortened,and the registration error is further reduced.The positioning accuracy of the hand-eye matrix obtained by the improved ICP algorithm for system three-dimensional positioning is 1.236mm.Compared with the traditional ICP algorithm,the measurement accuracy of the improved ICP algorithm for system three-dimensional positioning is 53.16%higher.It is proved that the improved ICP algorithm in this paper is of practical significance for improving the 3d positioning accuracy of the robot line-structured light vision guidance system. |
PDF Full Text Request