製品概要
概要
The FP-SNS-STAIOTCFT function pack for STM32CubeMX enables AI-based inertial applications on STM32 microcontrollers.
It supports the STEVAL-MKBOXPRO, STEVAL-STWINBX1, and STEVAL-STWINKT1B development boards, offering three main functionalities: AI inference on the MCU, machine learning core (MLC), and intelligent sensor processing unit (ISPU).
The pack includes sample applications for vibration monitoring and smart asset tracking, utilizing sensors like the ISM330DHCX and LSM6DSV16X.
The FP-SNS-STAIOTCFT seamlessly integrates with STM32CubeMX, allowing easy configuration and code generation.
Key software packs required include X-CUBE-AI, X-CUBE-ISPU, and X-CUBE-MEMS1, ensuring comprehensive support for AI, sensor interfacing, and preprocessing.
The included AI smart sensor monitoring (ai_ssm) firmware is a power-saving application designed for the STEVAL-MKBOXPRO that uses the MLC to classify motion patterns and transmit results via Bluetooth® LE.
The MCU is kept in low-power mode and is only interrupted by the smart sensor when an event is detected.
Two operation modes are available: connection mode and beacon mode.
The connection mode is activated at boot or by pressing User button 1 or triggering NFC. In this mode, a blue LED lights up, and the board is connectable, ready to receive a new MLC model from a client (like the ST AIoT Craft mobile app or gateway).
The beacon mode is entered after client disconnection and reboot. The board operates in low-power, broadcasting AI prediction results as Bluetooth® LE beacons (advertisement frames) when an event occurs or after a 1-minute keepalive timeout. An orange LED briefly flashes upon interrupt/beacon transmission.
The goal of both operation mode is to provide smart AI sensing with flexible Bluetooth® LE communication and robust power management for extended battery life, making it ideal for scalable IoT applications.
-
特徴
- AI inertial:
- AI inference capabilities: it supports AI inference directly on the MCU, MLC, and ISPU, enabling applications such as vibration monitoring and smart asset tracking
- Seamless integration: it is compatible with STM32CubeMX for easy project configuration and code generation, supporting STEVAL-MKBOXPRO, STEVAL-STWINBX1, and STEVAL-STWINKT1B boards
- Comprehensive sensor support: it utilizes sensors such as ISM330DHCX and LSM6DSV16X for inertial data gathering and processing
- Multifunctional application: it offers ready-to-use projects with functionalities like AI inference, USB connectivity, and PnPL commands for switching between applications
- Robust software pack selection: it requires X-CUBE-AI, X-CUBE-ISPU, and X-CUBE-MEMS1 for AI, sensor interfacing, and preprocessing support
- Example of data logging firmware included in the function pack
- AI SSM:
- The AI smart sensor monitoring firmware utilizes the machine learning core (MLC) on the SensorTile.Box pro for power-efficient motion pattern classification and Bluetooth® LE communication
- Key features and operation modes:
- Objective: to achieve power saving by monitoring motion patterns directly in the smart sensor (using MLC) while keeping the microcontroller (MCU) in a low-power mode
- Initial state (Boot): the board is placed in Bluetooth® LE connectable mode. It waits to receive an MLC model from a client device (for example, a gateway or mobile app)
- Model deployment/post-disconnection: upon the client disconnecting, the board reboots. It enters a low-power mode. It then awaits prediction events (interrupts) from the smart sensor
- Event handling and data transmission: when a motion event is detected, the sensor interrupts the MCU. The firmware wakes and reads the MLC prediction data. The result is then transmitted immediately using Bluetooth® LE beacon mode (advertisement frames)
- Keepalive mechanism: if no event occurs, the prediction data is still read and transmitted after a 1-minute keepalive timeout
- Compatible with STM32CubeMX and uses already available software packs like: X-CUB-BLEMGR, X-CUBE-NFC4/7, and X-CUBE-MEMS1
- AI inertial: