STMicroelectronics Web tools with smart sensors to accelerate AIoT project landing

     ST has launched a new web-based tool, ST AIoT Craft, that simplifies the development of node-to-cloud AIoT (Artificial Intelligence for the Internet of Things) projects and associated network configurations on the machine learning kernel (MLC) of ST's intelligent MEMS sensors. The MLC machine learning kernel is a unique feature of the ST MEMS portfolio that enables sensors to run decision tree learning models directly. MLC is capable of operating autonomously, without the involvement of the host system, with low latency, low power consumption, and efficient processing of tasks requiring AI capabilities, such as classification and pattern detection.

     ST AIoT Craft also integrates all the steps needed to implement the development configuration of iot projects using MLC in sensors, and provides developers with a secure and easy-to-use development method to provide strong security protection for cloud data. This is an online tool that runs on a web browser and is very easy to use without downloading software. Because there is no need to download and install, it saves users a lot of time and makes it easier for teams (for example, AI experts and embedded software engineers) to work together.

     In this software, the AutoML function is a tool for creating a decision tree model that automatically selects the best properties, filters, and window sizes for sensor datasets. The framework can also train decision trees to run on MLC and generate configuration files to deploy the trained models. For beginners, this software is a quick start guide to getting an initial understanding of ST smart sensors and simplifies the development of AI applications. In addition, when network connectivity for iot projects needs to be configured, gateways can be set up with the DSM tool (Data Adequacy Module). The DSM module intelligently filters the data points to be transmitted to the cloud, optimizes the amount of communication data, minimizes power consumption, and facilitates future secondary training.

     Users can find code examples on how to build iot sensor-to-cloud solutions using decision trees, including fan coil monitoring, asset tracking, human activity recognition, and head pose. The samples can be downloaded and installed on ST's iot reference boards, such as SensorTile.box Pro, STWIN, and Stwin.box, and developers can run routine reviews, customize the samples, import their own data, or refine existing data sets to speed up project development.

     ST AIoT Craft is included in the ST Edge AI Suite package. This package provides developers with all the software tools, examples, and models they need to develop and deploy machine learning algorithms on ST's edge AI products. St's edge-AI edge intelligence products include STM32 microcontrollers (MCUS), Stellar MCUS and MEMS sensors with built-in MLCS or Smart Sensor processing units (ISpus).