计算机工程与应用 ›› 2019, Vol. 55 ›› Issue (15): 235-240.DOI: 10.3778/j.issn.1002-8331.1810-0357

• 工程与应用 • 上一篇    下一篇

三维立体视觉机械臂智能抓取分类系统的开发

张建民,许志辉,龙佳乐,陈富健,罗顺祺,罗鑫春,林根源,李鸿彬   

  1. 1.五邑大学 智能制造学部,广东 江门 529020
    2.深圳大学 信息工程学院,广东 深圳 518060
  • 出版日期:2019-08-01 发布日期:2019-07-26

Development of Intelligent Grasp and Classification System for Robot Arm with Three-Dimensional Vision

ZHANG Jianmin, XU Zhihui, LONG Jiale, CHEN Fujian, LUO Shunqi, LUO Xinchun, LIN Genyuan1, LI Hongbin   

  1. 1.Smart Manufacturing Department, Wuyi University, Jiangmen, Guangdong 529020, China
    2.School of Information Engineering, Shenzhen University, Shenzhen, Guangdong 518060, China
  • Online:2019-08-01 Published:2019-07-26

摘要: 采用工业相机、工业投影机、普通摄像头、计算机和机械臂开发了一套具有三维立体视觉的机械臂智能抓取分类系统。该系统采用自编软件实现了对工业相机、工业投影机的自动控制和同步,通过前期研究提出的双波长条纹投影三维形貌测量法获取了物体的高度信息,结合opencv技术和普通摄像头获取的物体二维平行面信息,实现了物体的自动识别和分类;利用串口通信协议,将上述处理后的数据传送至机械臂,系统进行几何姿态解算,实现了智能抓取,并能根据抓手上压力反馈自动调节抓手张合程度,实现自适应抓取。经实验证明该系统能通过自带的快速三维形貌获取装置实现准确、快速的抓取工作范围内的任意形状的物体并实现智能分类。

关键词: 三维形貌测量, 机械臂, 条纹投影, 几何姿态解算, 智能抓取

Abstract: An intelligent grasp and classification system of manipulator with three-dimensional stereo vision is developed by using industrial camera, industrial projector, ordinary camera, computer and robot arm. The system realizes the automatic control and synchronization of industrial camera and industrial projector by using self-compiled software. The height information of object is obtained by the method of double-wavelength fringe projection three-dimensional topography measurement proposed in the previous research. The two-dimensional parallel plane information of object is obtained by combining opencv technology and ordinary camera, and the automatic recognition of objects is realized. The data are transmitted to the manipulator by serial communication protocol. The intelligent grasp is realized by calculating the geometric attitude of the manipulator. The grasping degree can be adjusted automatically according to the pressure feedback on the gripper. Experiments prove that the system can capture objects of arbitrary shape within the working range accurately and quickly, and can classify the objetcts intelligently by its own fast 3D shape acquisition device.

Key words: 3D shape measurement, robot arm, fringe projection, geometric attitude calculation, intelligent grasp