STMicroelectronics has expanded its NanoEdge AI Studio machine-learning (ML) design software with support for smart sensors that contain ST’s embedded Intelligent Sensor Processing Unit (ISPU). The new release extends the market-unique tool’s capability to enable on-device learning of artificial intelligence (AI) models for anomaly detection inside intelligent sensors.

Distributes On-Device Learning

Designers can now use NanoEdge AI Studio to distribute inference workloads across multiple devices. This includes microcontrollers (MCUs) and sensors with ISPUs in their systems, significantly reducing application power consumption. Always-on sensors that contain the ISPU can perform event detection at very low power, only waking the MCU when the sensor detects anomalies.

End-to-End Automated Workflow

The tool provides a complete end-to-end automated workflow that significantly eases development of high-performing AI algorithms. Cases in point are anomaly detection, classification, and regression. On-device learning also permits development without requiring an exhaustive dataset to manage pre-deployment training; thus, increasing convenience and efficiency. In addition, support for incremental learning adds extra flexibility to complete partially trained models.

The libraries generated by NanoEdge AI Studio can run on any STM32 MCU, from entry-level devices containing the Arm Cortex-M0 core to the high-performance MCUs containing the Cortex-M7 core. It also newly supports ISPU-enhanced sensor.  This includes the recently announced ISM330ISN 6-axis inertial measurement unit (IMU).

The latest version of NanoEdge AI Studio delivers unique support for embedded AI development. Specifically, it combines dedicated hardware, software, tools, sample code, support, and training for developers.