SparkFun Releases Open-Source Machine Learning Platform

June 25, 2019 by Scott McMahan

SparkFun released the company's first open-source, embedded-systems module, SparkFun Artemis, Engineering Version. The SparkFun Artemis is intended to empower engineers, prototype makers, and R&D teams to integrate the TensorFlow machine-learning platform into any design.

Additionally, the SparkFun team has launched three boards with the unshielded module: BlackBoard Artemis; BlackBoard Artemis Nano; and BlackBoard Artemis ATP.

The Artemis Module measures 15.5mm x 10.5mm including an antenna. Primary features of the ultra-low-power Artemis module include a cortex-M4F based BLE module using the Apollo3 microcontroller from Ambiq. An advanced HAL (hardware abstraction layer) allows users to take advantage of the Cortex-M4F architecture. An integrated Bluetooth 5 low-energy radio and a 2.4GHz antenna are also included. All the necessary circuitry for easy integration such as large SMD pads and spacing allow for low-cost 2-layer carrier board implementations.

Related: Aspinity Enables Machine Learning with Analog Processor

The module has the capability of running machine learning algorithms with the low current consumption of 6µA/MHz at 3.3V. It can be programmed over pre-configured serial bootloader or JTAG. The device is also camera capable. It is equipped with a ISO7816 Secure 'Smart Card' interface, secure firmware update system, flexible serial peripherals, and it has a rich set of clock sources.

"SparkFun has always been about making emerging technologies available to a wider audience—the Artemis module is no exception," said SparkFun CEO Glenn Samala, "Our goal is to blow the doors open for innovation in the area of machine learning with our open-source-hardware model—the Artemis module moves us several steps in the right direction."

The three carrier boards provide somewhat familiar form factors and enable anyone to flex the capabilities of Artemis module.

Features of All Three Boards

A digital MEMS microphone allows for experimenting with always-on voice commands with TensorFlow and machine learning.

The carrier boards have built-in BLE radio and are fully compatible with SparkFun's Arduino core. They can be programmed easily under the Arduino IDE (SparkFun wrote the core from scratch).

Three boards with the unshielded module: BlackBoard Artemis; BlackBoard Artemis Nano; and BlackBoard Artemis ATP

An exposed JTAG connector lets the more advanced users who prefer to use the power, flexibility, and speed of professional tools. Qwiic® connectors enable easy prototyping and daisy-chain capabilities for the Qwiic ecosystem, which includes 70+ sensors, outputs, and controllers.

The BlackBoard Artemis comes in the Uno footprint and the company says it is a good entry point for working with the Artemis module.

Of the three boards, the BlackBoard Artemis Nano is the smallest form factor and provides excellent flexibility for smaller projects. The Blackboard Artemis ATP ("all the pins") breaks out all 48 GPIO of the module and allows users to leverage the full capabilities of the module.

SparkFun says it is seeking FCC/CE approval for the Artemis module which it will release as shielded modules, and the company will make it will available in tape-and-reel quantities. The newly released Engineering Version of the Artemis module lets users get started as soon as possible.