A Recognition System of AR Markers Attached to Carts in a Factory

In this paper, we propose a system for recognizing carts by attaching AR markers to them. In recent years, the Internet of Things (IoT) has been introduced into factories in the manufacturing industry. However, factories that produce a wide variety of products in small quantities still use carts for...

Full description

Saved in:
Bibliographic Details
Published in2022 IEEE/SICE International Symposium on System Integration (SII) pp. 608 - 613
Main Authors Kitsukawa, Takumi, Takahashi, Masahiro, Moro, Alessandro, Harada, Yoshihiro, Nishikawa, Hideo, Noguchi, Minori, Hamaya, Akifumi, Umeda, Kazunori
Format Conference Proceeding
LanguageEnglish
Published IEEE 09.01.2022
Subjects
Automation
Cameras
Deep learning
Image recognition
Manufacturing industries
Production facilities
System integration
Online AccessGet full text

Cover

More Information
Summary:In this paper, we propose a system for recognizing carts by attaching AR markers to them. In recent years, the Internet of Things (IoT) has been introduced into factories in the manufacturing industry. However, factories that produce a wide variety of products in small quantities still use carts for their operations, and automation has not progressed. Therefore, a method of attaching a low-cost AR marker to the cart and using a fixed-point camera to recognize the ID is considered. When using this method, it is necessary to improve the recognition performance of the marker by using image processing because the marker attached to the cart is small. In the proposed system, markers are detected using an object-detection method based on deep learning in images acquired by a fixed-point camera and recognized by a combination of cropping, preprocessing, and deblurring. As a result, the distance from which AR markers can be recognized increased from 2.9 m to 3.9 m. The recognition rate was improved from 12% to 81% with a distance of 1 m and a speed of 0.25 m/s. It has also been confirmed that the system can be processed online. We verified the practicality of the system by conducting an experiment using a cart in an actual factory.
ISSN:2474-2325
DOI:10.1109/SII52469.2022.9708838