XaLogic K210 AI Accelerator Looks to Bring Kendryte K210-Powered AI Acceleration to the Raspberry Pi
Compact add-on offers 0.5 TOPS of compute performance and the promise of open source software with full examples.
Following the success of the Kendryte K210 system-on-chip (SoC) for edge AI processing, XaLogic is looking to bring the same technology to the Raspberry Pi family of single-board computers (SBCs) — using a simple add-on HAT.
"K210 AI Accelerator is a Raspberry PI HAT powered by the Kendryte K210 AI Processor," the company explains of the design. "It fits into any [Raspberry Pi] with a standard 40 pins connector and gives it a Neural Engine Accelerator (Kendryte K210) with 0.5 TOPs (Tera Operations) processing power. Using one of our many free pre-trained models, you can add machine vision features using deep learning in a matter of minutes."
"This handy HAT, for example, lets you add AI features to your [Raspberry Pi] based camera even if you don’t know how to train your model. Our plugin module together with pre-trained models will make your camera AI-enabled within minutes."
Kendryte's K210 has proven popular for edge AI tasks, from dedicated computer vision boards to USB-connected accelerators — not to mention single-board computers. It's also not the first time XaLogic has looked to hook one up to a Raspberry Pi: Last year the company launched the XAPIZ3500 in a near-identical form factor.
The redesigned HAT, adhering to the Hardware Attached on Top (HAT) standard and using the same footprint as the compact Raspberry Pi Zero family of single-board computers, will come with open source software, pre-trained models, sample Caffe and TensorFlow projects, and secure connectivity to Amazon Web Services through a built-in Infineon Trust-M security module.
More details on the board, and for a form to fill in order to be alerted when the crowdfunding campaign for the board goes live, can be found over on Crowd Supply.