The Low-Power AI Module for Battery-Powered Applications

Image by Unsplash

This post is also available in: עברית (Hebrew)

Ambient Scientific has introduced a groundbreaking AI module, the Sparsh board, designed to revolutionize edge AI applications by operating on a coin cell battery. This innovation promises to bring artificial intelligence to a range of devices that require long-lasting, low-power operation, making it ideal for applications like human activity recognition, voice control, and wearables.

One of the most notable features of the Sparsh board is its ability to run continuously for months on a single coin cell battery, minimizing the need for frequent replacements. This is made possible by the integration of Ambient Scientific’s ultra-low power GPX10 AI processor, which utilizes advanced analog in-memory computing technology to deliver efficient performance without compromising on AI capabilities. According to Interesting Engineering, the module includes a variety of components, such as motion sensors, a digital microphone, and a Bluetooth Low-Energy (BLE) module, providing a robust platform for developers to quickly prototype battery-operated AI applications.

Ambient Scientific’s GPX10 processor is specifically designed for power-sensitive embedded AI tasks, making it an ideal solution for always-on devices. It features ten DigAn AI cores, a system-on-chip (SoC) design, and an ultra-low power analog-to-digital converter (ADC), all contributing to its ability to execute neural network functions while optimizing power consumption. This architecture allows developers to create flexible AI applications for a wide range of use cases, from wearables to voice-controlled devices.

To support developers, Ambient Scientific gave access to sample applications and algorithms, making it easier to begin building AI-powered projects right away.

The Sparsh board’s potential spans numerous sectors, from AI-enabled medical devices to predictive maintenance and even voice-controlled toys. Ambient Scientific plans to showcase the Sparsh AI module at CES 2025, offering live demonstrations of its capabilities, such as fall detection and voice recognition, to highlight its applicability for both professionals and hobbyists looking to innovate with battery-efficient AI solutions.