Pick and Place of Large Object Based on 3D Vision

Wu Hsien-Huang, Xie Jia-Kun

doi:10.5954/icarob.2020.os18-6

Automation is a necessary tool to achieve unmanned factory, and machine vision plays a vital role for providing intelligent recognition in automation. In this study, technique of 3D camera is used for 3D image capturing and matching to identify, pick and place large objects automatically. The system uses a commercially available 3D stereo vision camera to build the image acquisition system, and reconstruct the large objects in 3D. This 3D image is used for classifying the objects with the 3D object recognition algorithm. After the object was identified and its 3d information was obtained, a robot arm integrated with the camera system can be used for grasping. Compared with traditional 2d image matching for 3d big object recognition, stable 2d image features are much harder to obtain due to the shadow. The 3D stereo vision camera does not require strict requirements for lighting control, and only needs stable ambient light. The reduction of the difficulty in building image acquisition environment for large 3d object and the cost of camera system provide a new option for applications that requires large object identification.

Pick and Place of Large Object Based on 3D Vision

Description

Journal

Keywords

Details 詳細情報について

Export

Report a problem