Representational image of coin cell battery. AlexLMX/iStock
Ambient Scientific has unveiled its new AI module, the Sparsh board, which operates on a coin cell battery, making it suitable for a wide array of on-device AI applications.
The module aims to offer solutions for tasks such as human activity recognition, voice control, and acoustic event detection.
This innovation is notable for its ability to function continuously for months without frequent battery replacements.
The Sparsh board has various components, including motion sensors, a digital microphone, a Bluetooth Low-Energy (BLE) module, and more.
This combination makes it a powerful tool for developers looking to prototype battery-operated AI applications quickly.
AI module
According to GP Singh, the Founder and CEO of Ambient Scientific, traditional microcontroller units (MCUs) often require a compromise between AI performance and power efficiency.
However, introducing their ultra-low power GPX10 AI processor marks a new direction in the industry, utilizing advanced analog in-memory computing technology.
This allows developers to create accurate AI applications without sacrificing battery life or performance, making the technology versatile and flexible for various uses.
Ambient Scientific also provides a comprehensive software stack that streamlines the development process for AI applications.
The stack supports well-known frameworks like TensorFlow and Keras alongside a proprietary compiler designed to accommodate nearly all key types of neural networks.
Additionally, developers can access sample AI applications and algorithms, allowing them to start building projects almost immediately after downloading the AmbiSense SDK.
The range of applications explored with the Sparsh board extends across multiple sectors, including predictive maintenance, AI-enabled medical devices, wearables, and voice-controlled toys.
As interest grows from product developers, students, and researchers, Ambient Scientific plans to release various reference designs for battery-driven AI applications, aiming to enhance the accessibility and affordability of AI technology.
Coin cell battery
The GPX10 processor is a top choice for always-on embedded AI applications, specifically for power-sensitive edge devices.
Featuring ten DigAn AI cores, an integrated system-on-chip (SoC) design that includes an ultra-low power analog-to-digital converter (ADC), custom SRAM, and support for various peripherals, the GPX10 is touted as a highly flexible solution for edge AI requirements.
The architecture of GPX10 is specifically crafted for executing neural network functions, optimizing both power consumption and programmability. This emphasis on design allows for greater adaptability in applications requiring on-device AI.
Ambient Scientific offers its Development Kit (DVK) to aid developers, providing a comprehensive platform for testing and fine-tuning embedded AI applications.
The accompanying software development kit includes tools that let users create AI applications utilizing popular frameworks while offering a compiler and training toolkit to simplify the development process.
RECOMMENDED ARTICLES
Ambient Scientific is set to showcase the Sparsh AI module at an upcoming event, promising live demonstrations of its capabilities, including applications focused on fall detection and voice recognition.
This launch marks a significant step forward in bringing efficient, battery-operated AI solutions to the market, further cementing Ambient Scientific’s commitment to advancing AI technology.
With these advancements, both hobbyists and professionals can look forward to new opportunities to develop innovative AI solutions that are both powerful and practical.
Ambient Scientific will showcase its Sparsh AI module at CES 2025 with live demonstrations of AI applications running on coin cell batteries, such as Fall Detection, voice recognition, and more.
