基于双目视觉的散乱工件识别定位技术研究

Research on Identification and Positioning Technology of Scattered Workpieces Based on Binocular Vision

工业生产中,工件散乱堆积在传送链上的现象愈发严重,使得机器人在定位抓取作业时难度增大,目前较多依靠人工完成散乱工件的放置与分装,降低了生产效率。为提高工厂效益,促进工业生产智能化的快速发展,分析双目视觉系统的组成及原理,介绍目前基于双目视觉的散乱工件识别定位的主要方法,包括基于图形或边缘特征、ICP点云配准以及机器学习的目标识别定位方法,并比较不同方法下的散乱工件识别定位评价,总结上述方法在工业生产中的实际应用情况,阐述该领域现阶段研究存在的主要问题,同时对未来研究进行展望。

In industrial production,the phenomenon of scattered and stacked workpieces on the transmission chain is becoming more and more serious,which makes it more difficult for the robot to locate and grab workpieces.At present,the placement and packing of scattered workpieces are mostly completed manually,which reduces the production efficiency.Therefore,in order to improve the efficiency of the factory and promote the rapid development of the intelligent industrial production,this paper first analyses the composition and the principle of binocular vision system,then introduces main methods of the scattered workpiece recognition and positioning based on binocular vision,including target identification methods based on graphics or edge characters,ICP point cloud registration and machine learning.The identification and positioning evaluation of scattered parts using different methods are compared,and the practical application of the above methods in industrial production is summarized.Finally,the main problems existing in this field are discussed and the future research is prospected.