A recent study published in the journal Manufacturing Letters explored how integrating edge computing with the Industrial Internet of Things (IIoT) and computer vision can improve industrial processes. The goal was to address the challenge of managing the large amounts of data generated by IIoT devices.
The researchers introduced a novel, cost-effective architecture to help manufacturers, particularly small and medium-sized enterprises (SMEs), adopt advanced technology.
This system digitizes analog gauge readings on manufacturing floors, thereby enhancing accuracy and operational efficiency. Furthermore, a case study demonstrated the framework’s effectiveness in real-world settings.
Industry 4.0: Advancement in Manufacturing Process
Industry 4.0 is transforming manufacturing by integrating technologies like the Internet of Things (IoT), cloud computing, edge computing, digital twins, and big data analytics. Within this framework, the IIoT allows seamless data communication among industrial devices, enabling real-time monitoring and control, which are crucial for maintaining competitiveness. However, managing and analyzing vast amounts of data remains a significant challenge.
Edge computing addresses these challenges by processing data closer to its source, reducing latency and bandwidth demands typically associated with cloud computing. Local data processing enables faster decision-making and greater operational efficiency.
Additionally, integrating computer vision with IoT devices automates tasks like reading analog gauges, previously done by humans, reducing errors and improving accuracy.
Edge Computing to Enhance IIoT Applications
In this paper, the authors developed software and hardware architecture that uses edge computing to enable computer vision applications on manufacturing floors. They aimed to create a cost-effective, scalable solution accessible to SMMs, who often face high costs when adopting advanced technologies.
The study used Raspberry Pi microcomputers and Logitech C270 USB webcams as IoT devices, along with a Python-based edge node server as the system’s central component. This setup collects and processes visual data from analog gauges, commonly used in manufacturing. Emphasizing open-source technology and affordable hardware, the proposed system offers an accessible option for manufacturers with limited resources.
The framework includes several key components: a connection service for IoT devices, a communication system for data collection, computer vision algorithms implemented using the OpenCV library, and a database for storing parameters.
Multithreading was employed to handle multiple connections simultaneously, ensuring efficient communication without data loss, and the MQTT protocol facilitated reliable data transfer.
The researchers conducted a test case study focusing on digitizing readings from analog gauges. This application was chosen because it is relevant to various manufacturing contexts, where accurate monitoring of process variables is crucial.
Key Findings and Insights
The outcomes indicated that the proposed edge-computing architecture significantly improved the performance of IIoT applications in manufacturing, outperforming non-edge-enabled approaches.
During experimental validation, the authors tested the accuracy of computer vision algorithms in reading analog gauges, achieving less than 2% deviation from actual values across different pressure levels. This high level of accuracy is important for reliable data collection and analysis in manufacturing processes.
Additionally, the study compared the performance of the edge node to that of running the computer vision algorithms directly on IoT devices. It showed that the edge node achieved a 15.5% increase in performance, with faster interarrival times for data processing. This improvement underscored the value of edge computing in high-throughput environments.
Scalability was further examined by connecting multiple cameras to a single Raspberry Pi microcontroller. The system efficiently managed multiple video streams, though performance became unstable beyond three cameras due to hardware limitations. These results highlight potential cost savings, as multiple devices can share a single processing unit, thereby reducing overall implementation costs for manufacturers.
Industrial Implications
This research has significant potential for various manufacturing applications. The proposed architecture can be utilized for real-time monitoring and automation of processes that heavily depend on analog gauges, such as pressure and temperature measurements.
By digitizing these readings, manufacturers can improve operational efficiency, reduce errors, and implement predictive maintenance strategies.
Additionally, integrating computer vision with IIoT facilitates advanced quality control, enabling manufacturers to detect defects and anomalies in real-time. This capability is essential for maintaining product quality and meeting industry standards. The proposed solution's accessibility also empowers SMMs to adopt advanced technologies.
Conclusion and Future Directions
In summary, the novel framework proved effective for supporting SMEs in enhancing their manufacturing processes. It addresses the challenges of massive data generation in IIoT applications while offering a scalable, cost-effective solution for SMMs.
The successful digitization of analog gauge readings demonstrates the system’s practical value, paving the way for broader adoption of edge-enabled IIoT solutions in manufacturing.
Future work should focus on refining the architecture, exploring additional hardware and software configurations, and developing standardized interfaces to integrate diverse computer vision algorithms. By advancing the accessibility and functionality of these technologies, this research shows the potential to drive further 4.0 advancements in manufacturing automation and productivity in the industry.
Journal Reference
Katsigiannis, M., & Mykoniatis, K. Enhancing industrial IoT with edge computing and computer vision: An analog gauge visual digitization approach. Manufacturing Letters, 2024, 1264-1273. DOI: 10.1016/j.mfglet.2024.09.153, https://www.sciencedirect.com/science/article/pii/S2213846324002372
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.