Features

• Desktop tool for design and generation of STM32-optimized libraries: anomaly

and outlier detection, feature classification, and extrapolation of temporal and

multivariable signals

– Anomaly detection libraries are designed using very small datasets: learn

normality directly on the STM32 microcontroller and detect defects in real

time

– One-class classification libraries for outlier detection, designed with a very

small dataset: acquisition during normal equipment operation and detection

of any abnormal pattern deviation

– N-class classification libraries designed with very small, labeled dataset:

classify signals in real time

– Extrapolation with small, fragmented dataset by means of regression

libraries: prediction of future values based on data patterns never seen

before

• Supports any type of sensor: vibration, magnetometer, current, voltage, multi-

axis accelerometer, temperature, acoustic and more

• Explore millions of possible algorithms to find the optimal library in terms of

accuracy, confidence, inference time, and memory footprint

•

Generate very small footprint libraries running down to the smallest Arm

Cortex

‑M0 microcontrollers

• Embedded emulator to test library performance live with an attached STM32

board or from test data files

• Native support for STM32 development boards, no configuration required

• Easy portability across the various STM32 microcontroller series

Product status link

NanoEdgeAIStudio

Automated Machine Learning (ML) tool for STM32 microcontrollers

NanoEdgeAIStudio

Data brief

DB4564 - Rev 3 - December 2021

For further information contact your local STMicroelectronics sales office.

www.st.com

1 Description

NanoEdge

™

AI Studio (NanoEdgeAIStudio) is a new Machine Learning (ML) technology to bring true innovation

easily to the end-users. In just a few steps, developers can create an optimal ML library for their project, based on

a minimal amount of data.

A demonstration version is available for three months for free experiments. A professional version provides the

yearly Solo or Team licenses for embedded developers.

To help users to bootstrap their projects, STMicroelectronics proposes the Edge AI Sprint Package to limit risks

and investments while increasing the chances of success. This is a bundle that includes training sessions, a

NanoEdge

™

AI Studio license, and technical support.

Check the ordering information section from the data brief for more details and contact STMicroelectronics sales

office or authorized business partners to proceed with an order.

NanoEdge

™

AI Studio, also called the Studio, is a PC-based push-button development studio for developers,

which runs on Windows

or Linux

Ubuntu

One of its big advantages is that NanoEdge

™

AI Studio requires no specific data science skills. Any software

developer using the Studio can create optimal ML libraries from its user-friendly environment with absolutely no

Artificial Intelligence (AI) skills.

The Studio can generate four types of libraries: anomaly detection, outlier detection, classification, and regression

libraries.

An anomaly detection library is generated from a minimal amount of data examples showing normal and

abnormal behaviors. Once created, load the library into the microcontroller to train and infer directly on the device.

The library learns the equipment behavior from data acquired locally and adapts to each equipment behavior.

Once trained, the library inference compares data coming from equipment over time against the locally created

models to identify and report anomalies.

Outlier detection can be used to detect any abnormality with the one-class classification method. No example

of abnormal behavior is needed. Import normal signals into the Studio and easily create an optimized outlier

detection ML library.

A classification library can be used to classify a collection of data, representing different types of equipment

defects (such as bearing problems, cavitation problems or others) or different types of events in equipment

environment. Import the signals into the Studio and, in just a few steps, create a classification ML library that

gathers all this knowledge into a single library. When running on the microcontroller, the classifier analyzes the

live data and indicates the percentage of similarity against this static knowledge.

A regression algorithm can be used to extrapolate data and predict future data patterns. Import signals and

targeted values in the desktop tool and generate in a few steps a smart library to, for example, improve energy

management or forecast the remaining lifetime of an equipment.

These libraries can be combined and chained: anomaly or outlier detection to detect a problem on the equipment,

classification to identify the source of the problem, and regression to extrapolate information and provide real

insight to the maintenance team.

The input signals can range from vibration to pressure, sound, magnetic, time of flight just to name a few, or even

a combination of several signals. Multiple sensors can be combined, either in a single library, or using multiple

libraries concurrently.

Both learning and inference are done directly inside the microcontroller by means of the NanoEdge

™

AI self-

learning library, which streamlines the AI process and significantly reduces development effort, cost and therefore

time to market.

NanoEdgeAIStudio

Description

DB4564 - Rev 3

page 2/7