ST AIoT Craft revolutionizes the way that you develop edge AI and IoT solutions, with a seamless approach to real-time or offline dataset creation and management, and generation of AI algorithms for the machine learning core (MLC) technology inside MEMS sensors.
Experience a no-code environment leveraging AutoML to effortlessly build machine learning models based on decision trees for inertial MEMS sensors. You can validate model performance directly on hardware sensor nodes via intuitive web and mobile apps, guaranteeing real-world reliability. ST also provides reference designs and components to help you explore and evaluate use cases and proofs-of-concept for your innovations, along with project and model management functionality to help you streamline development and reduce time-to-market.
Edge AI classification results can be used in end-to-end applications thanks to the ability of ST AIoT Craft to deploy and configure sensor nodes, gateways, and cloud data ingestion resources.
The latest ST AIoT Craft 1.1 release includes enhancements for sensor AI programming, data privacy, MLC configuration files, and smart MEMS data logging.
ST AIoT Craft is nominated as finalist in the Artificial Intelligence category for Embedded World 2025 event. (embedded world | embedded award)
Enhanced embedded AI programming in MEMS sensors
ST AIoT Craft rev1.1 features the MLC expert mode to enhance embedded AI programming capabilities in smart MEMS sensors. This advanced mode allows for sophisticated machine learning configuration and customization directly through the cloud tool, enabling superior inference capabilities and cutting-edge AI functionality in your ultralow power applications.
Privacy and security
ST AIoT Craft rev 1.1 lets users deploy IoT data ingestion resources in their Azure cloud account, providing data control and system observability. By adopting existing user security protocols, this capability not only minimizes the risk of data breaches and unauthorized access, but also ensures compliance with stringent company data protection regulations.
Standardization for MLC configuration file
The ST AIoT Craft 1.1 tool for MEMS sensors uses the widely implemented JSON format for the MLC configuration file. JSON is a popular text-based data format that is easily read by humans and understood by computers. Its language-independent nature makes it ideal for exchanging data across different programming languages and platforms.
Extended data logging capability
The support of ST AIoT Craft 1.1 for Datalog2 v.3.0.0 extends the data logging capabilities for multiple ST MEMS sensors, providing users with enhanced features and greater flexibility in data collection management. These advancements ensure that developers can capture, tag, store, and analyze data more effectively, driving better insights and decision making.
What are the key features of ST AIoT Craft 1.1 version?
- Enhanced embedded AI programming in sensors (MLC expert mode). Benefits: advanced customization of the decision tree classifier in the MEMS sensor, and improved algorithm accuracy.
- Privacy and security (deploy IoT systems in user Azure cloud accounts). Benefits: enhanced control, increased security, and regulatory compliance from the developer/customer perspective.
- Standardization for MLC configuration file format (to JSON format). Benefits: easy to use and human and computer readable.
- Extended data logging capability for MEMS sensors (support of Datalog2 v.3.0.0). Benefits: Advanced data logging features and comprehensive data management.
ST AIoT Craft is the first online tool designed to program ST MEMS sensors embedding a machine learning core (MLC) for inference operations, making Artificial Intelligence of Things (AIoT) accessible to all. This tool enables developers with even limited experience to build projects in minimal time.
We provide precompiled application examples, such as asset tracking, human activity recognition, and head gestures. The tool also features an ML auto function to generate MLC configuration files automatically, with options for customization in the new MLC expert mode. This allows programmers with limited AIoT experience to observe the code and gain practical knowledge, and experts to experiment deeply with ST smart sensor functionality.
For more experienced users, the tool can deploy an end-to-end system, including flashing a Raspberry Pi to act as a gateway that collects data and connects to a cloud instance, aiding data logging and remote control of nodes. The enhanced data logging functionalities and standardized format ensures a seamless experience in data collection, which is the first important step in the ML journey.
ST AIoT Craft addresses the challenges around the democratization of AIoT. It enables both novices and experts to work on the same platform to rapidly build and deploy AIoT applications on sensors with a machine learning core. Users now have the freedom to deploy the development in their own Azure account, which no other tool on the market today offers. This high degree of accessibility and functionality in ST MEMS sensors is crucial for popularizing AIoT. Running decision trees on a sensor offers significant benefits, including major energy savings and enhanced privacy, as data remains on the sensor.
Additional resources:
- Documentation ST AIoT Craft
- Product overview
- Introduction video
- Product presentation
- Blog post "ST AIoT Craft, the first cloud solution for ST’s Machine Learning Core makes edge AI and IoT more accessible"
First published on Mar 6, 2025
