Workpiece Intelligent Identification and Positioning System based on Binocular Machine Vision

In this paper, a workpiece detection and positioning system has been studied. Firstly, data sets of workpieces with three different shapes (cube, cylinder and sphere) are established, and the YOLOV3 target detection algorithm is used for deep learning to realize intelligent recognition of different shapes of workpieces. Then by the use of binocular machine vision technology, the key points position of workpieces can be successfully detected with relative error less than 2 %. This workpiece detection and positioning system can be introduced in the mechanical arm grabbing control system to make it have the visual ability similar to human in order to realize mechanical arm intelligent grabbing in complex environment.