Smart factories: Optimizing operations, safety, and productivity with 5G and automation

This post has been republished via RSS; it originally appeared at: New blog articles in Microsoft Community Hub.

This one-pager outlines a use case for industries and manufacturing as, for example, consumer electronics manufacturing, but it can be perfectly applied to almost all domains that involve factories (such as clothing, as an example).

Nowadays, every manufacturer applies some quality and security controls inside the factories. Those can be divided into people security (employees) and quality control (what is being manufactured). Those two vectors can be automated in a relatively easy way by controlling what the manufacturer considers most important constantly, as “anything can happen anytime”.

Potential solution’s goals:

• For quality control thermal video images can be analyzed searching for defects, or microscopical images looking for toxic fibers.
• For security control, it is possible to address face recognition or Individual Protection Equipment (IPE) detection, among others.

The main idea of this use case is to improve the quality and the security of the factory by creating a 5G network connected to a Multi Edge Computing (MEC) solution, which can handle all those event detections and sends the images and the alerts wherever the manufacturer wants in order to take action, all of this in a fraction of a second. In an industry where a failure can stop production for hours, those detections are key to success.

Opportunities and goals

The main goal of this use case is to increase the quality of the factory’s outcome, creating an automated quality control with minimum human engagement.

The other goal that some factories may want is to increase the security of their employees by controlling automatically that they comply with all the safety requirements that the factory has. As an example, controlling if an employee does not wear a helmet is a relatively easy task, or virtual fencing some dangerous zones. This is also done automatically.

The outcome of all of this is a slight reduction of human workforce, the disappearance of human error due to fatigue or human limitations, and the possibility of detecting anomalies in a fraction of a second in a system that works 24 hours a day if needed.

“It is time to modernize factories in order to improve the quality and safety. Data driven results and alerts improve chain production and productivity, while increasing the output’s quality”

The Solution

What: Capture live video streaming via cameras (CCTV cameras, thermal if needed) and send that video in high quality through a 5G network for Edge devices to analyze the images.

Some of the main features of the solution are:

• Ultra-bandwidth: Up to 4K simultaneous video processing with no lag, and up to thousands of videos can be sent at the same time
• Outdoor scenarios: Outdoor scenarios completely possible, like video processing, due to the high capacity 5G has in outdoor environments. Interesting for video surveillance, as an example.

Another possibility that arises is that the system is IoT capable:  IoT scenarios are a match for the 5G network, as the lag is measured in milliseconds and the number of devices that can be connected is measured in millions. This enables to “sensorize” the machines and recreate a digital twin using the same network, to help gathering data that can prevent failures or double check quality control.

How: Powered by a Private 5G Network at the factory, with an edge compute in house, latency remains minimal while band goes up to Gbps. Combining this backbone with high quality cameras which can be 5G (and therefore, wireless) can obtain a major advantage as no physical connections have to be maintained or even hidden to avoid accidents or malicious manipulations.

Why: Because 5G is the “new” network standard and factories can take a major advantage when using this state-of-the-art technology. Having a Private 5G network enables many use cases to be developed in the factory, with almost no limits. Embracing 5G + AI means lower detection failure rate, faster event detection and less human cost.

Simplified Architecture

The Private 5G MEC solution has a relatively easy architecture that can be deployed in a relatively short period of time. The architecture is the following:

It is possible to see that the architecture has been designed for camera processing (quality and security), but also keeping in mind possible upgrades with IoT sensors inside the factory. With a local 5G core, managing the network, and edge compute, very low latency event detections are achieved.

Finally, the cloud component is used for three main processes:

• Data: Saving in databases (as Azure Data Explorer), Digital Twins, batch processing
• AI: Model training, model management, MLOps
• Device Management: Management of the Edge Computes and the sensors

This architecture is resilient and expandable because of its nature, two characteristics every factory needs in their operations.

Request to download the technical guide to learn more about the design and architecture. 

It’s time to get started with 5G

The vision for smart roads is here, and Microsoft has resources that can help you get started on building modern connected applications using 5G, edge computing, and the cloud today. You can use the links below to learn more about smart roads and other use cases, access developer resources, or connect with a Microsoft representative.