Bosch BHI260AP self-learning AI smart sensor

Bosch’s BHI260AP, a self-learning AI sensor for wearables and hearables, integrates a six-axis inertial measurement unit (IMU), a 32-bit customer-programmable microcontroller and software functionalities in a system-in-package (SiP) solution. It is an all-in-one unit for always-on sensor applications, suitable for wrist wearables, hearables, smartphones, AR/VR headsets and controller devices.

The smart sensor features embedded edge artificial intelligence (AI) with self-learning AI software, including fitness tracking, navigation (pedestrian dead reckoning), machine learning analytics, and relative and absolute orientation estimation. The integrated, six-degrees-of-freedom IMU includes a 16-bit, three-axis accelerometer and 16-bit, three-axis gyroscope. The host interface is configurable as SPI or I²C, with two master interfaces (one selectable SPI/I²C and one I²C) and up to 12 GPIOs.

The smart sensor’s self-learning and personalisation features enable users to easily train devices with customised fitness activities. The sensor includes more than 15 pre-learned fitness activities, so no training is required before use. In addition to the included activities, the device also offers the ability to learn, personalise, auto-track and enhance the user experience. Users can also enhance or add more fitness activities without modifying the software or requiring an original dataset.

Since the AI is running on the sensor itself, connecting to the cloud or a smartphone is not required. This innovative edge AI functionality is designed to reduce latency, minimise power consumption and improve user privacy.