Aetina/Millitronic Partnership Results in a Smarter Factory

March 19, 2019 Alice Lai, Aetina Corp.

Factories of the future must adhere to the Industry 4.0 standards to ensure a high level of automation, robots that work in a cohesive manner, and a faster time to market for the end products. To achieve these goals, such factories must lower their latencies, increase their connection rates, provide a method to maintain and control the robotics, and offer the flexibility needed to maintain and upgrade the equipment when necessary.

A good starting point toward such a factory is the use of Edge computing and Wi-Fi that’s based on the latest 5G standard. Edge computing, a staple of the Industry 4.0 factory, enables the distribution of computing resources across the network, and in a perfect world, using dynamic allocation of those resources. As such, data can be processed as close as possible to where it is input. This (hopefully) eliminates the need to transfer data to the cloud, where latencies and potentially significant costs can be introduced.

To prove the point, let’s take a quick look at an intelligent automated textile factory that makes use of Edge computing and 5G Wi-Fi. Textiles have come a long way since their early days thanks to things like green technologies and the addition of smart electronics. That complexity allows manufacturers to keep end-product defects to a minimum. In many factories, defect detection is a manual process and certainly not a perfect one. Automation takes the inspection and detection process to levels previously unseen.

What’s needed to implement Industry 4.0 scenarios in such a factory? You need an intelligent Edge computing platform with high data-transfer rate, zero-latency video inspection with multiple high-speed cameras, and a fixed-wire access (FWA) appliance on the textile inspection machine. This will minimize inspection time while maximizing yields.

Because there are lots of textile types available, the platform must be smart enough to adapt, while offering an upgrade path as new technologies and standards become available. Thanks to machine vision, some form of machine learning can also be applied to continually update the process. This could either be performed at the Edge, or if significant number crunching is required, it can be done in the Cloud.

When you throw in the multi I/O requirements, like up to six cameras, and space limitations of the application, a platform like Aetina’s AN310, using an nVidia Jetson fits the bill. This platform can embed Millitronic's 5G WiGig hub solution using a USB 3.0 adaptor to deliver wireless Gigabit Ethernet data transfers for the first- and last-mile 5G applications.

While the benefits that the combined Aetina and Millitronic solution offer to the textile space are obvious, those same attributes can just as easily be applied to any factory that requires some form of computer vision and/or machine learning. The potential benefits include more flexible I/O, a smaller form factor compared to competitive offerings, and reduced human interaction, ultimately leading to fewer errors and a faster manufacturing process.

The Aetina AN310 Edge AI computing platform employs an nVidia Jetson TX2/TX1 module. Each module is powered by 256 nVidia CUDA cores. The AN310 supports a wide temperature range (-40°C to +85°C). For even tighter areas, there’s the credit-card-sized AN510, which can also handle a Jetson TX2/TX1.

The Millitronic WiGig hub is suited for unlicensed 60-GHz millimeter-wave (mmW) networks. It provides high capacity, security, low latency, and extends a local-area or mesh network without interference from 2.4/5-GHz Wi-Fi and LTE networks.

Alice Lai is a representative of Aetina Corp.

Previous Article
Five Minutes With…Michael Randt, Founder/CEO, Trinamic
Five Minutes With…Michael Randt, Founder/CEO, Trinamic

The space where the electronics meets the mechanics ends up being a place of disconnect. Who owns that doma...

Next Article
DevKit for inference at the edge
DevKit for inference at the edge

AI development kit with integrated Linux Ubuntu oeprating system, SDKs and pretrained models.

How to Develop Cross-Industry IoT Interoperability

Multi-Part Series