Unlock embedded AI potential

 

We bring you deep learning data processing
at sub-milliwatt scale.

Discover COLIBRY NPU

ASYGN Colibry Neural Processing Unit (NPU) is an ultra-low power microcontroller designed to accelerate AI-driven processing for imagesound, or any sensors data. It integrates a fully reconfigurable and efficient Neural Network Processing Accelerator (NNPA), enabling high-performance and ultra-low power solutions for a wide range of industrial applications.​

Ultra-low power Microcontroller

32-bit RISC-V

Up to 300 MHz

<1mW – 10mW

128 KB SRAM

Neural Network Accelerator

On-chip NN inferences

Fully reconfigurable

Up to 9.6 GOPS

640KB SRAM

Sensor data processing

Parallel camera interface

Audio and SPI/I2C motion support

On-chip FTT and ISP accelerator

TinyML ready

Optimized for frugal AI

Enabling always-on AI

Best-in-class power efficiency

Applications

Warehouses and factories

Detection of abnormal
machinery noise

Food tracking in a large
refrigerated storehouse

Healthcare

Fall detection

Vital state monitoring

Agriculture

Early identification of crop infection

Pest monitoring for optimize treatment

Sport

Steps counting

Posture and movement validation

Security

Gun fire detection

Suspicious activity recognition in restricted area

Key Features

< 100 mW
AI processing
0 GOPs
Compute power
0 MHz
Max frequency
0 -bit
RISC-V CPU
0 kB
CPU SRAM
0 kB
NNPA SRAM

Our chip in your system

Image processing

From object detection to scene analysis, the Colibry NPU enables multi-FPS inferences on images at sub-milliwatt scale.

Sound sensing

Sounds classification, noise cancellation, sources identification… The colibry NPU is the perfect chip to build tiny and high-accuracy sound-based systems.

Motion classification

Ideal for wearables tracking physical activity and security systems detecting motion, building frugal AI-enabled motion sensors becomes possible.

Discover our GREENAI initiative

An energy-autonomous European industrial platform for frugal and embedded AI