Dual-camera-based method for identification and location of scattered self-plugging rivets for robot grasping. (February 2019)

• Record Type: Journal Article
• Authors: Jiang, Tao; Cheng, Xiaosheng; Cui, Haihua; Shi, Cheng; Li, Yufei
• Is Part Of: Measurement. Volume 134(2019)
• Page Start: 688
• Page End: 697
• Publication Date: 2019-02
• Subjects: Binocular vision -- Self-plugging rivet -- Identification and location -- Robot grasping; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2018.11.017
• Languages: English
• ISSNs: 0263-2241

Abstract

Highlights: Effective combination of morphology technique enhance the image features. Mask image subtraction and multiple operation of it clarify an unique rivet pose. Dual-camera technique ensures the accurate location of scattered rivets. The proposed system makes the riveting process more flexible. Abstract: With the development of robotic vision technology, applications of visual guided robotic grasping and assembly become more and more widespread in aerospace industry. Self-plugging rivet with complex structures plays an important role in connection of aircraft and spacecraft. This paper focus on identification and location of scattered self-plugging rivets for robot automatic grasping. In this paper, the structure and rivet measurement elements are analyzed and the calculation method of grasping point and angle is introduced, firstly. Secondly, the rivet image processing method based on multiple corrosion and expansion operators is proposed, and rivet contour is obtained initially. Thirdly, in order to determine the grasping angle uniquely, a method of multiple times subtraction with mask image is proposed. In addition, the rivets' skeleton lines are rebuilt with binocular and the effects of threshold value are studied. The reconstruction and gripping accuracy is verified through a standard board with marker points. Vision-guided robot rivet grasping experiment succeed finally. Experiment results show our detection algorithm is effective and the grasping system can completely satisfy the industrial requirements.