The Bristol-based AI specialist has built a package of tools and resources to accelerate the development of devices that can ‘think’ for themselves. The new XMOS SDK – which includes libraries to support FreeRTOS operation on xcore.ai, and TensorFlowLite for Microcontroller models – will enable developers to rapidly deploy custom or off-the-shelf AI models using a standard framework alongside all of the control, communications, signal and I/O processing required to create a complete and secure application solution.
As XMOS explains, the kit equips developers with standardized tools and resources that they need to create devices that absorb contextual data from their environment, infer meaning from that data, and translate the results into action.
The SDK includes AIoT tools, with scripts, tools and libraries to convert TensorFlowLite for Microcontroller models into a format that targets accelerated operations on the xcore.ai platform; libraries to support FreeRTOS operation on xcore.ai, providing a familiar, standard industry programming environment, and allowing developers to work in examples showing a variety of operations based on bare-metal and FreeRTOS operation, including smart microphone sensing.
Documentation includes getting started guides, example builds and execution walkthroughs, as well as access to XMOS’ open-source libraries of interfaces & signal processing algorithms. These tools will enable developers to rapidly deploy custom or off-the-shelf AI models using a standard framework alongside all of the control, communications, signal and I/O processing required to create a complete and secure application solution.
"Our AIoT SDK enables developers to create intelligent endpoint-AI solutions for a huge variety of applications," says Mark Lippett, CEO of XMOS. "The flexibility of the xcore.ai architecture enables our customers to create truly differentiated solutions using standard embedded software techniques like TensorFlowLite for MCU in a fraction of the time required using traditional hardware approaches."
Early access to the XMOS AIoT SDK will be available on the GitHub open source platform, designed to be used in conjunction with the xcore.ai Explorer Kit, which is available on limited release via xmos.ai.
Future releases will include other xcore.ai hardware platforms, targeting specific use case applications. This includes a smart home platform – a small form-factor reference design with additional Wi-Fi capability, designed to demonstrate the capabilities of voice at the edge of networks and due to be released in early 2021.