| 论文简介 |
Modern industrial manufacturing has more intelligent requirements for industrial robots, such as intelligent grasping, industrial parts assembling, in which the robot should measure the pose of a part before manipulating it. In this paper, a multi-view-based 6-DOF pose estimation method and visual guidance system is proposed. First, we setup a visual guidance system for robotic intelligent grasping, which can estimate the 6-degree-of-freedom pose of the part and guide the robot to grasp it accurately. Second, we introduce a fast hand-eye calibration method that can quickly calibrate the relative transformation between the camera and the robot end-effector. Third, we propose a 6-degree-of-freedom pose estimation method based on multiple views for large size parts. Compared with existing methods, the present method can estimate the 6-DOF pose of large-size parts through capturing large-size parts from several different photograph positions and orientations. Specially, we use binocular reconstruction technology to calculate the initial value of the pose before nonlinear optimization, to ensure the validity and accuracy of the final pose. Experimental tests demonstrated our proposed method can obtain accurate 6-DOF pose and the visual guidance system can accomplish intelligent grasping tasks. |
(创新港)
