A vision-based robotic grasping system using deep learning for garbage sorting

• Published in: Chinese Control Conference pp. 11223 - 11226
• Main Authors: Chen Zhihong, Zou Hebin, Wang Yanbo, Liang Binyan, Liao Yu
• Format: Conference Proceeding
• Language: English
• Published: Technical Committee on Control Theory, CAA 01.07.2017
• Subjects: Cameras; Complex backgrounds; Deep Learning; Garbage sorting; Grasping; Machine vision; Manipulators; Object recognition; Robot kinematics; Robotic grasping; Sorting
• ISSN: 1934-1768
• DOI: 10.23919/ChiCC.2017.8029147

This paper proposes a robotic grasping system for automatically sorting garbage based on machine vision. This system achieves the identification and positioning of target objects in complex background before using manipulator to automatically grab the sorting objects. The object identification in complex background is the key problem that machine vision algorithm is trying to solve. This paper uses the deep learning method to achieve the authenticity identification of target object in complex background. In order to achieve the accurate grabbing of target object, we apply the Region Proposal Generation (RPN) and the VGG-16 model for object recognition and pose estimation. The machine vision system sends the information of the geometric centre coordinates and the angle of the long side of the target object to the manipulator which completes the classification and grabbing of the target object. The results of sorting experiment of the bottles in the garbage show that the vision algorithm and the manipulator control method of the proposed system can achieve the garbage sorting efficiently.