Embedded World 2023: Nordic Show us Matter, Wifi 6 Location Tracking and EdgeAI

Nordic Semiconductor had a range of hardware demos on display at Embedded World 2023, covering everything from Smart Home device control to IoT and embedded machine learning. The new nRF7002 WiFi 6 IC featured prominently, and we saw demos of how it slots into systems running the new Matter protocol, along with new ways to track location with a low power cost.

We started by chatting with Finn Boetius, Product Marketing Engineer at Nordic, about their Matter Smart Home Demo featuring the Thingy:53, nRF7002 DK, and Smart Home products from Apple, Google and Eve.

The demo highlighted just how powerful Matter is when compared to how closed Smart Home tech was in the past. Not only does the system incorporate both WiFi and Thread connectivity, but it shows how competing Smart Home devices can now work seamlessly in the same network.

We also saw a demo of how the nRF7002 allows for low-power WiFi-based positioning, courtesy of an as-yet-unreleased extension board featuring the WiFi 6 IC, with Arduino shield-compatible pins. It was running alongside an nRF9160 DK and several Thingy:91s, communicating via cellular IoT and providing GNSS/GPS positioning along with the new cell-based WiFi location features.

There's no release date for the nRF7002 EK extension board, but according to the staff at the Nordic booth, it will be coming this year!

Finally, Robin Maristad Saltnes, Nordic Radiographer & electrical engineer, showed us just how easy it can be to train and deploy EdgeAI models on the Thingy:53 prototyping kit using Edge Impulse.

Despite a couple of technical hitches (due to a conference floor full of radio interference and one attendee who decided to stand right in from of the camera at one point), it was incredible to see the workflow in action. With just a computer - or even a mobile phone - you can capture sensor data from the accelerometer, microphone, or environmental sensors, and upload it to Edge Impulse to train a model in the cloud. The model can then be deployed straight back onto the device and used. The entire process is quick and requires no coding knowledge. It's incredible to see, given how hard any kind of Machine Learning was a very short time ago.

To find out more about any of the products featured in these videos, head to the Nordic Semiconductor website.