Robotic manipulation via the assisted 3D point cloud of an object in the bin-picking application

Abstract

In the field of robot control, manipulation or object handling is one of the most critical tasks. The existing techniques reveal some challenges such as the unstructured nature of objects or their random orientations within cluttered environments. Our method has emerged as a promising solution, providing detailed spatial information that enhances object detection and pose estimation in this study. Initially, several mechanical computations are carried out to indicate the user-defined tool of robotic end-effector. Then, the image processing techniques, for instance HSV filter, are deployed to identify the center of target object. After that, the coordinates of an object can be obtained using the 3D point cloud data. This information is transmitted to our embedded computer via TCP/IP communication protocol. The outcome of the proposed approach is to properly enable the grasping operation without the human intervention. From these results, it can be seen obviously that our approach is feasible and can be applied in many industrial fields.