Programmable smart IMU features on-sensor AI-powered analysis

Mouser Electronics, Inc. now stocks the BHI385 smart IMU from Bosch. The device is an intelligent sensor system that provides accurate acceleration measurement up to ±28g, enabling precise motion analysis for high-impact events across sports, fitness, and gaming. The sensor combines a gyroscope and accelerometer with embedded AI software and an integrated 32-bit MPU, creating a complete programmable system based on Bosch Sensortec's field-proven IMU platform.

The sensor's embedded AI supports key motion functions, including 6DoF/9DoF sensor fusion, step counting, activity recognition, and gesture detection (single/double/triple tap). Self-learning cyclic gesture recognition enables it to detect repeated motion patterns and adapt them to the individual, recognising what is being done and how well it is being performed. This enables advanced use cases, such as evaluating the quality of movement in sports or rehabilitation training.

All processing runs on the integrated low-power processor, avoiding wake-up calls to the host CPU. Typical current consumption is less than 500μA at 50Hz for sensor fusion, and under 50μA for simpler accelerometer-based algorithms.

The device is offered in a compact 20-pin LGA package measuring 2.5mm x 3mm x 0.95mm³. Low power consumption and embedded intelligence make it an excellent fit for wearables, hearables, and edge AI systems.

The BHI385 IMU is supported by the BHI385 Shuttle Board 3.0, also available from Mouser. Facilitating the evaluation of the BHI385 smart sensor, the Shuttle Board 3.0 also includes a BME688 AI-based environmental sensor, a BMM350 magnetometer, and an onboard pressure sensor. The board consists of essential passive components and supplies access to all sensor pins, making it excellent for standalone testing and integration into larger development systems. The Shuttle Board is designed to work with the Bosch Application Board 3.1 sensor prototyping platform. The company's Motion AI Studio supports a fast time-to-market, while the Open SDK enables the integration of custom machine learning models.