STM tools speed edge AI projects to market

  • February 8, 2023
  • Steve Rogerson

Swiss electronics company ST Microelectronics has introduced set of tools and services to get edge AI technology on the market faster and with less complexity by aiding hardware and software decision-making.

The STM32 Cube AI developer cloud opens access to a suite of online development tools built around the STM32 family of microcontrollers (MCUs).

“Our goal is to deliver the best hardware, software and services to meet the challenges faced by embedded developers and data scientists so they can develop their edge AI application faster and with less hassle,” said Ricardo De Sa Earp, STM executive vice president. “We are unveiling the world’s first MCU AI developer cloud, which works hand-in-glove with our STM32Cube AI ecosystem. This new tool brings the possibility to remotely benchmark models on STM32 hardware through the cloud to save on workload and cost.”

Serving the growing demand for edge AI-based systems, the STM32Cube AI desktop front-end includes the resources for developers to validate and generate optimised STM32 AI libraries from trained neural networks. This is now complemented by the STM32Cube AI developer cloud, an online version of the tool.

It includes an online interface to generate optimised C-code for STM32 microcontrollers, without requiring prior software installation. Data scientists and developers can benefit from the STM32Cube AI’s neural network optimisation performance to develop edge-AI projects.

There is access to the STM32 model zoo, a repository of trainable deep-learning models and demos to speed application development. At launch, available use cases include human motion sensing for activity recognition and tracking, computer vision for image classification or object detection, and audio event detection for audio classification. Hosted on GitHub, these enable the automatic generation of getting-started packages optimised for the STM32.

And there is access to an online benchmarking service for edge-AI neural networks on STM32 boards. The cloud-accessible board farm features a range of STM32 boards, refreshed regularly, allowing data scientists and developers to measure the actual performance of the optimised models remotely.

The tool has been undergoing testing and evaluation by several embedded developers.

“We have used STM32Cube AI in the past with great success,” said Toly Kotlarsky from Zebra Technologies’ technical staff. “It has allowed us to implement high-performing AI applications running on low-cost MCUs. Today we are glad to see that this product is further evolving by offering an online interface. This will allow us to evaluate performance of the AI models and choose a proper hardware architecture earlier in the process so we can converge more quickly on the development of AI applications.”

Didier Pellegrin, vice president at Schneider Electric, added: “The model zoo, STM32Cube AI online interface and remote benchmarking capabilities on STM32 boards make it easier for our data scientists with various hardware knowledge to evaluate embeddability of AI models into our products’ microcontrollers. Additionally, being capable of testing our models on several STM32 microcontrollers in a few clicks enables us to consider embedded AI processing at an earlier stage in the design process and to take advantage of it to design advanced features.”

ST Microelectronics employs more than 50,000 creators and makers of semiconductor technologies. An integrated device manufacturer, it works with more than 200,000 users and thousands of partners to design and build products and ecosystems.