STMicro's Machine Learning-Enabled Motion Sensor Improves Accuracy, Cuts Energy
May 02, 2019
A machine learning core in the sensor works with integrated finite-state machine logic to classify motion data based on known patterns.
GENEVA. STMicroelectronics has integrated machine learning capabilities into its LSM6DSOX iNEMO advanced inertial sensors to improve battery life and performance of mobile devices and wearables. A machine learning core in the sensor works with integrated finite-state machine logic to classify motion data based on known patterns. By offloading these functions from a host processor, motion-based apps can be accelerated and energy conserved.
The LSM6DSOX contains a 3D MEMS accelerometer and 3D MEMS gyroscope, as well as substantial internal memory and high-speed I3C digital interface. The I3C interface enables short, high volume connections with the host controller and shorter overall connection times. The result is always-on performance and a low typical current consumption of 0.55 mA that helps maximize battery life.
The LSM6DSOX can be integrated with Android and iOS platforms, and users can train the machine learning core for decision tree-based classification using the open-source Weka application. The PC-based Weka app allows developers to generate settings and limits from sample data such as acceleration, magnetic angle, and speed to characterize various types of movement.
The LSM6DSOX is available now for $2.50 in qty 1000. More information is available at www.st.com/content/st_com/en/products/mems-and-sensors/inemo-inertial-modules/lsm6dsox.html.