We stopped at the Silicon Labs booth at Embedded World 2022 to talk over some of their latest hardware, and their Bluetooth 5 antenna array immediately caught our eye, along with a machine vision project using edgeAI deployed using Edge Impulse.
The Bluetooth tracking demo - dubbed the "Angle of Arrival demo" - uses Bluetooth 5.1 along with 4 BG22-RB4191A antenna arrays to sense the position of multiple BG22 asset tags in a 3d space. We chatted about the benefits of a system like this, along with how the new hardware can achieve sub-1-meter accuracies even in highly populated areas.
What the video doesn't show is that the tracking is indeed happening in 3d space. Soon after filming, the display showing position was flipped to an isometric view, showing the tag's height alongside floor position.
The BG22-RB4191A is available in the Silicon Labs online store, with regular unit for $149, and a Pro Kit including an extra mainboard and debugging adapter for $299.
Big Machine Vision, Tiny MCU
Just opposite, we saw a working example of a new MCU in the EFR32 line, this time incorporating super-low power on-device object recognition. We spoke with Tamas from Silicon Labs and Matt from Edge Impulse about the new setup.
The core of this project is the EFRMG24 series of wireless SoCs. Alongside a wide range of wireless connectivity options including high-performance 2.4 GHz RF and integrations such as Zigbee among others, it contains a dedicated "AI/ML hardware accelerator" responsible for the high-speed performance. More information on the whole range of EFR32MG24 Series 2 Multiprotocol Wireless SoCs can be found on the product page.
Perhaps unsurprisingly, Edge Impulse announced full support for the xG24 line of SoCs back in April, so if you are wanting to get started, you'll find a lot of out-of-the-box support for the latest and smartest of the EFR32 bunch!