Welding Seam Recognition Robots Based on Edge Computing
Author:
- 1 August 2020
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 27-30
Abstract
In order to meet the requirements of the accuracy and real-time performance during the working process of underwater welding robots, a scheme of welding seam recognition robots system based on the edge computing is proposed in this paper. A number of pre-processing methods for capturing welding seam image were designed, including Thresholding, Filtering and Edge Detect. A Convolutional Neural Network(CNN) model for welding seam recognition was also created. In the experiments, the image pre-processing and CNN algorithms were integrated in and deployed to the robots, and the learning and training algorithms of the CNN were deployed to the cloud servers. The image pre-processing methods filtered the interference in underwater operations and achieved the image compression and feature extraction. The cloud servers fulfilled the training and parameter optimization of the CNN, which improved the accuracy of welding seam image recognition.Keywords
This publication has 2 references indexed in Scilit:
- Learning Deep Architectures for AIFoundations and Trends in Machine Learning, 2009
- Gradient-based learning applied to document recognitionProceedings of the IEEE, 1998