Integrating Tiny Machine Learning and Edge Computing for Real-Time Object Recognition in Industrial Robotic Arms

By integrating visual recognition technology and multi-object recognition into robotic arms, the flexibility and automation of the production process were improved in this study. By applying tiny machine learning (TinyML) and machine vision algorithms, we integrated edge computing devices to control...

Full description

Saved in:
Bibliographic Details
Main Authors: Nian-Ze Hu, Bo-An Lin, Yen-Yu Wu, Hao-Lun Huang, You-Xin Lin, Chih-Chen Lin, Po-Han Lu
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Engineering Proceedings
Subjects:
Online Access:https://www.mdpi.com/2673-4591/92/1/74
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849471935949307904
author Nian-Ze Hu
Bo-An Lin
Yen-Yu Wu
Hao-Lun Huang
You-Xin Lin
Chih-Chen Lin
Po-Han Lu
author_facet Nian-Ze Hu
Bo-An Lin
Yen-Yu Wu
Hao-Lun Huang
You-Xin Lin
Chih-Chen Lin
Po-Han Lu
author_sort Nian-Ze Hu
collection DOAJ
description By integrating visual recognition technology and multi-object recognition into robotic arms, the flexibility and automation of the production process were improved in this study. By applying tiny machine learning (TinyML) and machine vision algorithms, we integrated edge computing devices to control the robotic arms and identified objects precisely on the production line, with ultra-low energy consumption. The developed system in this study included the SparkFun Edge development board and Raspberry Pi Camera Module 3, as edge devices for data processing, image recognition, and robotic arm control. By utilizing the Edge Impulse platform for data collection, model training, and optimization, edge devices and models for use in resource-limited environments were successfully generated. Using Edge Impulse’s automated toolchain, real-time image processing and object recognition were realized. The system achieved improved recognition accuracy and operational speed, demonstrating the potential of TinyML in enhancing the intelligence of robotic arms. MariaDB was chosen for data storage. Grafana was used to design a user-friendly web interface for real-time data monitoring and visualization and immediate data analysis and system monitoring. The developed system presented a success rate of 99% during actual operation. The feasibility of combining advanced image processing technology with robotic arms in intelligent manufacturing was verified in this study. The potential of integrating machine learning and automation technologies was also confirmed for the development of future manufacturing technologies. The model provides a technical reference and ideas for future factories that require high levels of automation and intelligence.
format Article
id doaj-art-e6717ef4e66447c2a664b41128d3777c
institution Kabale University
issn 2673-4591
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Engineering Proceedings
spelling doaj-art-e6717ef4e66447c2a664b41128d3777c2025-08-20T03:24:39ZengMDPI AGEngineering Proceedings2673-45912025-05-019217410.3390/engproc2025092074Integrating Tiny Machine Learning and Edge Computing for Real-Time Object Recognition in Industrial Robotic ArmsNian-Ze Hu0Bo-An Lin1Yen-Yu Wu2Hao-Lun Huang3You-Xin Lin4Chih-Chen Lin5Po-Han Lu6Smart Machinery and Intelligent Manufacturing Research Center, National Formosa University, Yunlin 63201, TaiwanDepartment of Information Management, National Formosa University, Yunlin 63201, TaiwanDepartment of Information Management, National Formosa University, Yunlin 63201, TaiwanDepartment of Information Management, National Formosa University, Yunlin 63201, TaiwanDepartment of Information Management, National Formosa University, Yunlin 63201, TaiwanDepartment of Information Management, National Formosa University, Yunlin 63201, TaiwanDepartment of Information Management, National Formosa University, Yunlin 63201, TaiwanBy integrating visual recognition technology and multi-object recognition into robotic arms, the flexibility and automation of the production process were improved in this study. By applying tiny machine learning (TinyML) and machine vision algorithms, we integrated edge computing devices to control the robotic arms and identified objects precisely on the production line, with ultra-low energy consumption. The developed system in this study included the SparkFun Edge development board and Raspberry Pi Camera Module 3, as edge devices for data processing, image recognition, and robotic arm control. By utilizing the Edge Impulse platform for data collection, model training, and optimization, edge devices and models for use in resource-limited environments were successfully generated. Using Edge Impulse’s automated toolchain, real-time image processing and object recognition were realized. The system achieved improved recognition accuracy and operational speed, demonstrating the potential of TinyML in enhancing the intelligence of robotic arms. MariaDB was chosen for data storage. Grafana was used to design a user-friendly web interface for real-time data monitoring and visualization and immediate data analysis and system monitoring. The developed system presented a success rate of 99% during actual operation. The feasibility of combining advanced image processing technology with robotic arms in intelligent manufacturing was verified in this study. The potential of integrating machine learning and automation technologies was also confirmed for the development of future manufacturing technologies. The model provides a technical reference and ideas for future factories that require high levels of automation and intelligence.https://www.mdpi.com/2673-4591/92/1/74Industry 4.0TinyMLmulti-object recognitionedge computingreal-time visualizationEdge Impulse
spellingShingle Nian-Ze Hu
Bo-An Lin
Yen-Yu Wu
Hao-Lun Huang
You-Xin Lin
Chih-Chen Lin
Po-Han Lu
Integrating Tiny Machine Learning and Edge Computing for Real-Time Object Recognition in Industrial Robotic Arms
Engineering Proceedings
Industry 4.0
TinyML
multi-object recognition
edge computing
real-time visualization
Edge Impulse
title Integrating Tiny Machine Learning and Edge Computing for Real-Time Object Recognition in Industrial Robotic Arms
title_full Integrating Tiny Machine Learning and Edge Computing for Real-Time Object Recognition in Industrial Robotic Arms
title_fullStr Integrating Tiny Machine Learning and Edge Computing for Real-Time Object Recognition in Industrial Robotic Arms
title_full_unstemmed Integrating Tiny Machine Learning and Edge Computing for Real-Time Object Recognition in Industrial Robotic Arms
title_short Integrating Tiny Machine Learning and Edge Computing for Real-Time Object Recognition in Industrial Robotic Arms
title_sort integrating tiny machine learning and edge computing for real time object recognition in industrial robotic arms
topic Industry 4.0
TinyML
multi-object recognition
edge computing
real-time visualization
Edge Impulse
url https://www.mdpi.com/2673-4591/92/1/74
work_keys_str_mv AT nianzehu integratingtinymachinelearningandedgecomputingforrealtimeobjectrecognitioninindustrialroboticarms
AT boanlin integratingtinymachinelearningandedgecomputingforrealtimeobjectrecognitioninindustrialroboticarms
AT yenyuwu integratingtinymachinelearningandedgecomputingforrealtimeobjectrecognitioninindustrialroboticarms
AT haolunhuang integratingtinymachinelearningandedgecomputingforrealtimeobjectrecognitioninindustrialroboticarms
AT youxinlin integratingtinymachinelearningandedgecomputingforrealtimeobjectrecognitioninindustrialroboticarms
AT chihchenlin integratingtinymachinelearningandedgecomputingforrealtimeobjectrecognitioninindustrialroboticarms
AT pohanlu integratingtinymachinelearningandedgecomputingforrealtimeobjectrecognitioninindustrialroboticarms