Anyone who uses a fitness tracker or a smartwatch is familiar with the problem: Many devices only know a limited number of activities or fail to reliably identify and track training due to different movement styles, devices and performance levels of the users. A new motion sensor with integrated AI functions for wearables should now help.

Thanks to the AI ​​functions Users can train the sensor in learning mode for new fitness activities or optimize their own movement sequences using the built-in personalization function. Manufacturers of wearables can easily expand the list of 15 activities already built into the sensor without having to program software. Automatic tracking should also work better thanks to AI.

Local AI without cloud New exercises from trainers, top athletes or friends can be transferred to the sensor to enable users to learn from and compare with others. The artificial intelligence runs locally as “edge AI” on a 15 – bit microprocessor in the sensor module, so learned activities can be carried out without an Internet connection or connection to an external one Application can be tracked and analyzed.

Bosch Sensortec supplies the KI sensor BHI 260 AP as an integrated system in package (SiP), consisting of hardware, software and the embedded AI. For this purpose, the manufacturer offers various special applications that the manufacturers of wearables can load onto the sensor. You can also program your own applications and load them onto the sensor.

(odi)