News

ST Releases Free AI Development Tool for All ST Cortex-M Devices

December 11, 2023 by Duane Benson

STMicroelectronics has announced new AI software libraries for unlimited deployment on any STM32 microcontroller—free of charge.

STMicroelectronics recently released its NanoEdge AI Studio free of charge for use on ST Cortex-M 32-bit microcontrollers (MCUs). NanoEdge Studio is a software package that provides machine learning (ML) and artificial intelligence (AI) libraries for its Arm Cortex-M microcontrollers.

 

NanoEdge Studio

NanoEdge Studio, now free, brings AI to Cortex-M 32-bit MCUs and beyond. 

 

The software allows designers to easily select and implement otherwise complex ML and AL libraries into C code. All About Circuits participated in a media briefing to learn more about the upgraded NanoEdge AI Studio. 

 

ST Helps Bring AI to More Edge Devices

Edge AI, like edge computing, takes place in edge-located and remote devices that have some level of processing power. Edge computing tends to work on smaller, specific tasks than centralized AI computing. However, due to the limited computing power in edge devices, edge AI is often more difficult. Code must be highly optimized for space and processing speed.

 

Facial recognition with NanoEdge libraries

Facial recognition with NanoEdge libraries.

 

This is where ST’s NanoEdge Studio comes into play. NanoEdge AI Studio has a library of specialized AI modules compatible with C code for Cortex-M microcontrollers. By using the studio’s library search capability, a software engineer can describe an application and choose from a set of libraries. This approach makes the use of AI practical in far more edge computing settings. Opening the studio up as a free offering makes deployment much more economical for various products.

"We believe that Nano AI Studio is so powerful that we decided to open it as widely as possible in an effort to accelerate the adoption of AI," said Remi El-Ouazzane, ST's president of MCUs and digital ICs group.

"Customers who would like to run their solution on a dual platform or even only on an Arm Cortex devices will be able to do so under a special license agreement."

 

NanoEdge Studio V4.3 Simplifies the Dev Process 

ST has been working on the core technology to put AI into microcontrollers for more than a decade. The company first ported AI models to existing STM32 MCUs, followed by ports to automotive MCUs. It then created NanoEdge Studio in 2021 to allow engineers to more easily deploy these ports. ST claims the newest version of NanoEdge Studio, V4.3, adds more automation to the development process and simplifies deployment. 

"Developers will find the right tools and resources in the ST Edge AI suite to achieve their goal easily and with optimized results," said Marco Cassis, ST's president of its analog, MEMS, and sensors group.

"We are keeping things simple: start with a single entry point (whatever ST device you are working on), use guided navigation through a wide choice of tools, and use the suite for any ST devices supporting STM32 MCUs and MPUs, MEMS, sensors, and a comprehensive collection of educational resources alongside an active community to help develop skills and implement technology more effectively."

 

NanoEdge development process diagram

NanoEdge development process diagram. 
 

Software developers can specify how much of the MCU’s resources can be devoted to the AI modules and then search the studio for an appropriate library. The library is linked into the developer’s C code—ready for deployment. The new version also improves validation to give developers confidence that the algorithms will operate as intended.

 

Free of Charge and Widely Applicable

AI programming is in high demand, but it is still a complicated and uncommon programming discipline. AI in the resource-constrained world of 32-bit microcontrollers is even more difficult. The NanoEdge Studio was created to bridge the capability gap and enable faster edge AI application deployment in more settings. Offering it for free will take away a major barrier to entry for companies deciding if the cost of edge AI development is worth the reward.

"[NanoEdge Studio] is designed to optimize and deploy neural networks on any ST devices," said El-Ouazzane. "It guarantees the best interaction and performance between software and hardware."

All libraries are compatible with any Arm Cortex-M STM32 microcontroller and are sensor agnostic. In addition, the libraries can be used with some non-ST Cortex-M microcontrollers, provided the chip vendor has a licensing arrangement with ST. With this compatibility, developers can work with custom products, 140 ST development boards, and more than 1,000 Cortex-M MCUs from other vendors.

"All the tools in this AI suite are completely free of charge for both development and production," El-Ouazzane reiterated. "That's part of our commitment to accelerate the adoption of AI technology. And as a further step in this commitment, we have, as a company, decided to make the deployment of the NanoEdge AI library free for STM32 and to make it available for any Cortex M-based MCUs under a special license agreement."

 


 

All images used courtesy of STMicroelectronics.