Open AccessReal-Time Multi-object Grasp Based on Convolutional Neural Network
Author(s) -
Hui Yu,
Yong Xu
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1631/1/012006
Subject(s) - grasp , artificial intelligence , robotic arm , computer science , object (grammar) , computer vision , robot , process (computing) , convolutional neural network , robot hand , position (finance) , finance , economics , programming language , operating system
The grasping of the robotic arm is the basic ability of the robot to perform a variety of complex tasks. For the robotic arm, visual-based grasping tasks are still a challenging problem. This paper presents a system to implement dynamic grasping of the robot arm. The neural network is used to detect objects in multiple scenes, and then the robot arm is controlled to achieve closed-loop grasp of the target object. In the process of grasping the target object, if the position of the target object moves, the robot arm will immediately stop and go to the new target position for grasping. The extensive experiments demonstrate that the proposed method enables the robotic arm to achieve dynamic grasping in real world.