Autonomous seam recognition and feature extraction for multi-pass welding based on laser stripe edge guidance network
- PDF / 1,847,531 Bytes
- 13 Pages / 595.276 x 790.866 pts Page_size
- 61 Downloads / 170 Views
ORIGINAL ARTICLE
Autonomous seam recognition and feature extraction for multi-pass welding based on laser stripe edge guidance network Kaixuan Wu 1,2 & Tianqi Wang 1,2
&
Junjie He 1,2 & Yang Liu 1,2 & Zhenwei Jia 1,2
Received: 28 July 2020 / Accepted: 13 October 2020 # Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract In this paper, an autonomous seam recognition and feature extraction method for multi-pass welding based on laser stripe edge guidance network is proposed to overcome the interference of strong reflection, spatter, and arc noise in actual welding environment. Firstly, the laser stripe edge guidance network consisting of modified VGGnet, progressive laser stripe feature extraction, non-local laser stripe edge feature extraction, one-to-one guidance module, and multi-feature fusion module is introduced to recognize the laser stripe under heavy arc noises. Afterwards, the gray centroid method is adopted to obtain the thinning laser stripe. Aiming at extracting the position of feature points, the least square method and non-uniform rational B-splines with second derivative are utilized. Finally, experiments and analysis show that our proposed method performs favorable in terms of effectiveness, flexible, accuracy, and robustness, which could meet the actual welding requirements. Besides, the maximum error and maximum root mean square error for feature extraction are 4.7 pixel and 1.78 pixel, respectively. Keywords Convolutional neutral network . Multi-pass seam recognition . Feature extraction . Curve fitting . Robotic welding . Laser vision
1 Introduction Multi-pass welding is widely used in the industrial manufacturing process of medium thick plates such as ships, bridge trusses, and pressure vessels [1]. At present, the teaching and playback method is usually adopted for each pass in actual industrial welding environment, which leads to high labor intensity and low efficiency. Moreover, the influence of thermal deformation in the welding process on the position of welding seam is ignored resulting in large welding error and poor quality [2]. In recent years, the laser vision–based welding robot technology is introduced to improve that situation and achieves favorable performance [3–9]. However, the obtained welding images for seam recognition and feature would be inevitably contaminated by the strong reflection, spatter, and arc noise as shown in
* Tianqi Wang [email protected] 1
Tianjin Key Laboratory of Advanced Mechatronics Equipment Technology, Tianjin 300387, China
2
School of Mechanical Engineering, Tiangong University, Tianjin 300387, China
Fig. 1 when implement the real-time seam tracking. Therefore, it is still hard to guarantee the accuracy and robustness in realtime multi-pass welding due to the interference of strong reflection, spatter, and arc noise. Many scholars have done a lot of works on seam recognition and feature extraction based on laser vison for multi-pass welding and proposed abundant effective and feasible methods. Cai et al. [10] introduced wavelet t
Data Loading...
