A novel robot welding path planning system for medium-thick plate based on 3D visual perception

Page view(s)
5
Checked on Sep 10, 2025
A novel robot welding path planning system for medium-thick plate based on 3D visual perception
Title:
A novel robot welding path planning system for medium-thick plate based on 3D visual perception
Journal Title:
Measurement
Publication Date:
24 March 2025
Citation:
Liu, Z., Li, J., Shi, C., Qin, L., Yang, S., Liu, N., & Shi, P. (2025). A novel robot welding path planning system for medium-thick plate based on 3D visual perception. Measurement, 252, 117365. https://doi.org/10.1016/j.measurement.2025.117365
Abstract:
To address inefficiency and inaccuracy in robot welding path planning for medium-thick plates, it is essential to move beyond the conventional methods of laborious teaching-playback and offline programming. To this end, this paper develops a novel robot welding path planning system for medium-thick plate based on 3D visual perception. Initially, aiming at accomplish the endpoint identification and output of positional information for the welding path of medium-thick plate, an improved structure-aware single-stage 3D object detection (SA-SSD) algorithm is implemented. Secondly, the improved algorithm integrates the TripletAttention module and the efficient pyramid squeeze attention network (EPSANet) module. Moreover, the minimum points distance intersection over union is applied to the 3D object detection scenario (3DMPDIoU) is designed for calculating the bounding box regression loss function. To address the challenge of generating welding paths for medium-thick plate workpieces with complex geometric structures. An improved cubic spline interpolation method is designed to fit the welding path. The experimental results confirm that path planning time of the proposed method is less than 22% of the time required by teaching-playback method, and the end coordinate error of each welding path is controlled within the 0.27mm error band. This method can achieve robotic welding path planning for medium-thick plate without teaching-playback and offline programming.
License type:
Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0)
Funding Info:
This work is supported by grants of the Chongqing Technology Innovation and Application Development Special Key Project (No: cstc2021jscx-gksbX0030), the Scientific and Technological Research Program of Chongqing Municipal Education Commission (No: KJQN202200826) and the Graduate Scientific Research and Innovation Foundation of Chongqing Technology and Business University (No: yjscxx2023-211-52).
Description:
ISSN:
0263-2241
Files uploaded:

File Size Format Action
revised-manuscript-with-no-changes-marked.pdf 3.13 MB PDF Request a copy