Seam tracking system based on laser vision and CGAN for robotic multi-layer and multi-pass MAG welding

Liu, Chenfan; Shen, Junqi; Hu, Shengsun; Dingyong, WU; Zhang, Chao; Yang, Hui

doi:10.1016/j.engappai.2022.105377

Cited by 16 publications

(3 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An attention dense convolutional block [16] is proposed to extract and accumulate multi-scale features. A restoration and extraction network [31], based on a conditional generative adversarial network, is also developed to extract the seam feature points. While these methods yield high accuracy performance, they use large models and lead to an increase in computational delay.…”

Section: Laser Stripe Extractionmentioning

confidence: 99%

DSNet: A dynamic squeeze network for real-time weld seam image segmentation

Chen,

Wang,

Shi

et al. 2024

Engineering Applications of Artificial Intelligence

View full text Add to dashboard Cite

Section: Laser Stripe Extractionmentioning

confidence: 99%

DSNet: A dynamic squeeze network for real-time weld seam image segmentation

Chen,

Wang,

Shi

et al. 2024

Engineering Applications of Artificial Intelligence

View full text Add to dashboard Cite

“…Among them, welding robots are the most widely distributed type of robot. Most of the current welding robots are based on demonstration type robots [1] . Although these types of robots can to some extent replace humans in completing the welding process, they also have prominent drawbacks.…”

Section: Introductionmentioning

confidence: 99%

Weld Image Processing Algorithm Based on Machine Vision Detection and Industrial Internet of Things

Zhu,

Wu,

Liu

2024

Preprint

View full text Add to dashboard Cite

In order to solve the problem of inaccurate centerline extraction caused by external environmental interference in traditional algorithms for processing weld seam images, which in turn affects the accuracy of robotic arm welding, the author proposes a weld seam image processing algorithm based on machine vision detection and industrial Internet of Things. This method first preprocesses the welding object image by grayscale, histogram equalization, and threshold segmentation. Then, the Canny operator is used to extract the pixel edge lines of the weld seam, and a dedicated centerline extraction algorithm for the weld seam is designed. Finally, an adaptive polynomial welding seam curve fitting algorithm was proposed, and the pixel coordinates of the fitting curve were converted into robot coordinates through a coordinate transformation model, in order to send welding path data to the arc welding robot. The experimental results show that the improved arc welding robot system can autonomously recognize complex weld seam trajectories and generate accurate welding paths, thereby guiding the robot to complete welding operations. The maximum error value is controlled within 0.29 millimeters, which meets the requirements of welding accuracy and shows good processing effect.

show abstract

“…For example, one study found that welding process variables affect the quality of robotic MIG welding of EN24T steel [5]. Another study developed a robotic seam-tracking system using conditional generative adversarial networks and laser vision (CGAN) [6]. Other study of robot welding application was of a robot-assisted MIG/MAG welding system called "MyWelder" which was proposed as an excellent productive, intuitive, and easily customizable technology [7].…”

Section: Introductionmentioning

confidence: 99%

Simulation of a Ship’s Block Panel Assembly Process: Optimizing Production Processes and Costs through Welding Robots

Wahidi,

Oterkus,

Oterkus

2023

JMSE

View full text Add to dashboard Cite

Conventional welding techniques for complex structures often rely on human involvement, which can be prone to errors when deviations from the planned process occur. In contrast, robotic welding is highly precise and effective, particularly in the assembly of complex structures such as double-bottom ships. Therefore, this paper presents a comprehensive technical and economic analysis comparing robotic welding to conventional welding in the assembly process of a ship’s block panels. The study aims to evaluate and compare the strategies employed in robotic welding and conventional welding, with a specific focus on the ship double-bottom context. Furthermore, an economic value analysis is conducted to assess the cost effectiveness of each approach. The analysis reveals that robotic welding can achieve a significantly faster welding speed, completing the process approximately 3.85 times quicker compared to conventional methods. Moreover, the ratio of electricity and man-hours between robot welding and conventional welding is 1:2.75. These findings highlight the potential for cost savings by implementing robotic welding processes. The analysis reveals a significant difference in operational costs, highlighting the efficiency and cost effectiveness of robotic welding compared to conventional methods.

show abstract

Seam tracking system based on laser vision and CGAN for robotic multi-layer and multi-pass MAG welding

Cited by 16 publications

References 25 publications

DSNet: A dynamic squeeze network for real-time weld seam image segmentation

DSNet: A dynamic squeeze network for real-time weld seam image segmentation

Weld Image Processing Algorithm Based on Machine Vision Detection and Industrial Internet of Things

Simulation of a Ship’s Block Panel Assembly Process: Optimizing Production Processes and Costs through Welding Robots

Contact Info

Product

Resources

About