Please use this identifier to cite or link to this item:
|Title:||Object recognition system for dual-arm robot in bin packing||Authors:||Yin, Wanqi||Keywords:||DRNTU::Engineering::Mechanical engineering::Robots||Issue Date:||2018||Abstract:||Most of the existing automated bin packing process is designed for packing standardized and identical items into the container without perception process. A perception process especially objects recognition system is able to extend the ability of the packing system to detect, recognize and localize the item and packs a set of random items with different sizes into the container efficiently. This report aims to develop an object detection and recognition system implemented on Kawada NEXTAGE Open, an industrial dual-arm robot for bin packing proposes in order to achieve the recognition and localization of the boxes for packing in logistic process. The report covers the introduction of the hardware platform and software environment setup, equipment usage, perception function development, system tests, performance result, and discussion. After reviewing the existing object recognition algorithm, a combination of the 3D point cloud and 2D RGB images was used as raw data for information extraction. 3D point cloud data was used for box recognition, coarse localization and pose identification. 2D RGB image was used for fine localization based on 2D recognition and coordinating the gripper pose with the box orientation. In this project, some tools were used including Point Cloud Library (PCL), OpenCV and all the functions developed using C++, and Python was implemented into Robot Operation System (ROS) together with the packing planning function and robot control function. The outcome of the system implementation as shown in the report, demonstrated that the desired task outcome was given, as the dual-arm robot was able to recognize and locate the boxes with various sizes and in various poses. The outcome successfully proves that the object recognition system enables the dual-arm robot to achieve a more intelligent performance in bin packing task.||URI:||http://hdl.handle.net/10356/75658||Rights:||Nanyang Technological University||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||MAE Student Reports (FYP/IA/PA/PI)|
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.