STMicroelectronics revealed its advanced inertial sensors integrated with machine-learning core for battery-friendly and improved activity-tracking. STMicroelectronics announced Motion Sensor with Machine Learning for the Indian market at Annual MEMS Media briefing held in India on 4 July which was addressed by Vishal Goyal, senior technical marketing manager – Analog and MEMS Group, RF, Sensors and Analog Custom Products, Asean-ANZ and India of STMicroelectronics.
The LSM6DSOX is a system-in-package featuring a 3D digital accelerometer and a 3D digital gyroscope boosting performance at 0.55 mA in high-performance mode and enabling always-on low-power features for an optimal motion experience for the consumer.
The LSM6DSOX delivers best-in-class motion sensing that can detect orientation and gestures in order to empower application developers and consumers with features and capabilities that are more sophisticated than simply orienting their devices to portrait and landscape mode.
The event-detection interrupts enable efficient and reliable motion tracking and contextual awareness, implementing hardware recognition of free-fall events, 6D orientation, click and double-click sensing, activity or inactivity, stationary/motion detection and wakeup events.
The LSM6DSOX supports main OS requirements, offering real, virtual and batch mode sensors. In addition, the LSM6DSOX can efficiently run the sensor-related features specified in Android, saving power and enabling faster reaction time. In particular, the LSM6DSOX has been designed to implement hardware features such as significant motion detection, stationary/motion detection, tilt, pedometer functions, timestamping
and to support the data acquisition of an external magnetometer.
The LSM6DSOX offers hardware flexibility to connect the pins with different mode connections to external sensors to expand functionalities such as adding a sensor hub, auxiliary SPI, etc.
The tilt function helps to detect activity change and has been implemented in hardware using only the accelerometer to achieve targets of both ultra-low power consumption and robustness during the short duration of dynamic accelerations.
The tilt function is based on a trigger of an event each time the device’s tilt changes and can be used with different scenarios, for example:
• Triggers when phone is in a front pants pocket and the user goes from sitting to standing or standing to sitting;
• Doesn’t trigger when phone is in a front pants pocket and the user is walking, running or going upstairs.
Significant Motion Detection
The Significant Motion Detection (SMD) function generates an interrupt when a ‘significant motion’, that could be due to a change in user location, is detected. In the LSM6DSOX device this function has been implemented in hardware using only the accelerometer.
SMD functionality can be used in location-based applications in order to receive a notification indicating when the user is changing location.
LSM6DSOX: Machine Learning Core
The Machine Learning Core (together with the Finite State Machine) is one of the main embedded features available in the LSM6DSOX. It is composed of a set of configurable parameters and decision trees able to implement algorithms in the sensor itself.
The LSM6DSOX embeds a dedicated core for machine learning processing that provides system flexibility, allowing some algorithms run in the application processor to be moved to the MEMS sensor with the advantage of consistent reduction in power consumption.
Machine Learning Core logic allows identifying if a data pattern (for example motion, pressure, temperature, magnetic data, etc.) matches a user-defined set of classes. Typical examples of applications could be activity detection like running, walking, driving, etc.
• Motion tracking and gesture detection
• Sensor hub
• Indoor navigation
• IoT and connected devices
• Smart power saving for handheld devices
• EIS and OIS for camera applications
• Vibration monitoring and compensation
Along with revolutionary sensor, ST demonstrated AlgoBuilder,Unico GUI & SensorTile.box.