STMicroelectronics LSM6DSOX: Machine‑Learning‑Enabled Inertial Sensor for Precise, Energy‑Efficient Activity Tracking

STMicroelectronics has embedded machine‑learning capabilities into its LSM6DSOX iNEMO sensor, elevating activity‑tracking accuracy while preserving battery life in mobile and wearable devices.

Featuring a dedicated machine‑learning core, the sensor classifies motion data against known patterns, freeing the main processor from this workload. This off‑loading reduces power draw and speeds up motion‑centric applications such as fitness logging, wellness monitoring, personal navigation, and fall detection.

Devices powered by the LSM6DSOX benefit from an “always‑on” experience without compromising runtime. The sensor’s larger internal memory and a cutting‑edge high‑speed I3C interface enable longer intervals between controller interactions and shorter connection bursts, delivering additional energy savings.

Integration with popular mobile platforms—Android and iOS—is straightforward, making the LSM6DSOX ideal for consumer, medical, and industrial markets.

The unit houses a 3D MEMS accelerometer and a 3D MEMS gyroscope, and its machine‑learning core processes complex movements while consuming a typical current of only 0.55 mA, further protecting battery life.

Its finite‑state machine logic works alongside the learning core to recognize motion patterns and detect vibrations. Developers can train the core with decision‑tree classifiers via Weka, an open‑source PC application, creating tailored settings that differentiate movements by acceleration, speed, and magnetic angle.

Additional features such as free‑fall, wake‑up, 6D/4D orientation, click and double‑click interrupts broaden application possibilities—from user‑interface controls to laptop protection and optical image stabilization (OIS) configurations.