Avnet Unveils the RASynBoard Edge AI Board, with Renesas MCU, Syntiant NDP, and TDK Sensors

Avnet has announced pre-orders for the RASynBoard, a development kit which combines a Renesas microcontroller and a Syntiant Neural Decision Processor (NDP) with on-board TDK sensors to provide a platform for ultra-low-power edge AI projects — with immediate support in Edge Impulse.

“Avnet’s strong relationship with industry leading suppliers such as Renesas, TDK, and our new supplier Syntiant allows us to combine the unique offerings and capabilities of each partner into a complete system-level, production-ready solution," explains Avnet's Jim Beneke of the new board. "Our design customization and manufacturing capabilities also allow us to offer production-ready solutions for customers needing a full-turnkey option."

The RASynBoard Core Board is a compact module with USB Type-C connectivity which includes a Renesas RA6M4 microcontroller with a single Arm Cortex-M33 core running at 200MHz, 256kB of static RAM, and 1MB of flash memory, a Syntiant NDP120 Neural Decision Processor (NDP) with Core 2 deep neural network hardware, Arm Cortex-M0 core, and Cadence HiFi 3 Digital Signal Processor (DSP). Elsewhere on the board is 2MB of SPI flash, a Renesas DA16600 802.11b/g/n Wi-Fi and Bluetooth 5.1 radio module, a lithium-polymer battery management circuit, and a six-axis inertial measurement unit (IMU) and digital microphone from TDK.

With all of the above in a compact 25×30mm (around 0.98×1.18") footprint, the RASynBoard aims to be a tiny titan for on-device machine learning work — loading models from the SPI flash storage for execution on the NDP120. "Our hardware technology brings advanced multimodal neural processing to Avnet's new RASynBoard with very low power consumption," explains Syntiant's Mallik Moturi. "The NDP120's ability to provide highly accurate, always-on sensor processing with relatively no impact to battery life was a key edge AI requirement in the development of the module, which also makes the device ideal for supporting a wide range of field applications in smart buildings, factories and cities."

For those growing beyond the RASynBoard Core Board's built-in capabilities, two 28-pin board-to-board connectors provide expansion — with an optional IO Board making use of these to offer an on-board debugger and USB-serial interface, a MikroE Click shuttle box header, a Pmod Type-6A socket, a 14-pin microcontroller expansion header, microSD storage, a user-definable button and RGB LED.

On the software front, Avnet has announced a partnership with Edge Impulse to provide support for the RASynBoard within the company's popular Studio machine learning platform. "Edge impulse is excited to collaborate with Avnet on the launch of the RASynBoard, an ideal solution for ultra-low-power machine learning applications thanks to its Renesas MCU and sub-mW Syntiant NDP120 processor, and flexible array of sensors," says Edge Impulse's Raul Vergara. "Customers can quickly develop advanced models in Edge Impulse Studio and deploy them to the board for always-on inferencing in almost any location or environment."

The RASynBoard Core Board and IO Board bundle is now up for pre-order on Avnet's site at $99, with delivery expected to take place late in the second quarter of 2023.