STMicroelectronics has introduced the LSM6DSV320X, a compact AI-enabled inertial measurement unit (IMU) that integrates a 3-axis digital gyroscope, a 3-axis low-g accelerometer (±16g), and a 3-axis high-g accelerometer (±320g) in a tiny 3 x 2.5mm package, ideal for space-constrained IoT devices such as wearables, smartphones, gaming controllers, smart tags, personal safety gear, and industrial monitoring systems.
Most motion sensors are designed to track daily activities or detect sudden impacts, but the LSM6DSV320X is built to do both. One accelerometer tracks everyday motion up to ±16g, while the other handles high-impact events up to ±320g. One of the most interesting features is its adaptive self-configuration (ASC), which allows real-time adjustment of internal settings based on activity without the need for a host CPU, optimizing power use and responsiveness. It also includes a machine learning core (MLC) capable of running up to eight decision trees and a finite state machine (FSM) to support local AI inference for context awareness (e.g. fall detection or sudden motion events).
STMicro LSM6DSV320X specifications:
- Motion Sensing
- Dual accelerometers
- Low-g – ±2/±4/±8/±16 g
- High-g – ±32/±64/±128/±256/±320 g
- Gyroscope – ±250/±500/±1000/±2000/±4000 dps
- Dual accelerometers
- Event Detection & AI Features
- Advanced pedometer, step detector, and step counter
- Motion events – Free-fall, wake-up, 6D/4D orientation, tilt, click/double-click, high-g shock
- Significant motion detection
- High-g peak tracking
- Tilt detection and false step rejection
- Interfaces
- I²C, SPI (3- and 4-wire), MIPI I3C v1.1
- Auxiliary SPI and I3C interface for OIS support
- Sensor hub support for up to 4 external sensors
- Performance
- Output Data Rate (ODR) – Up to 7.68 kHz for gyroscope and accelerometers
- Noise
- Low-g accelerometer: 60 µg/√Hz (HP mode)
- High-g accelerometer – 1000 µg/√Hz (HP mode)
- Gyroscope – 3.8 mdps/√Hz
- Misc
- Quad-channel architecture – UI, OIS, EIS, and high-g acceleration
- Embedded Machine Learning Core (MLC)
- Up to 8 decision trees, 16 results per tree
- Supports external sensor data via the sensor hub
- Programmable Finite State Machine (FSM) – 8 programmable state machines
- Adaptive Self-Configuration (ASC) for real-time self-tuning
- Embedded Sensor Fusion Low Power (SFLP) algorithm for orientation
- Embedded 4.5 KB Smart FIFO with compression
- Embedded 16-bit temperature sensor (±15°C offset, 256 LSB/°C)
- Power Management
- Supply Voltage – 1.08V to 3.6V
- Power Consumption
- 6-axis combo (gyro + low-g accel) – 0.67 mA
- 9-axis combo – 0.80 mA
- Low-power mode (LPM1) – as low as 4.5 µA at 1.875 Hz
- Package – LGA-14L, 3.0 x 2.5 x 0.83 mm
- Temperature range – -40°C to +85°C
- Compliant – ECOPACK and RoHS
Developers can use MEMS Studio and the ST AIoT Craft graphical interface to configure the sensor, build machine learning models, and deploy them directly to the chip. The sensor’s Machine Learning Core (MLC) supports up to eight parallel decision trees, while an integrated Finite State Machine (FSM) handles complex motion detection scenarios locally. The company also offers the Motion XLF library to fuse data from both accelerometers, enhancing signal clarity and reliability for high-impact or precision use cases.
One of the most unique features of this IMU is its pedometer. The pedometer uses a four-stage process to count a person’s steps by detecting motion. It first computes the acceleration magnitude for orientation-independent detection, applies an FIR filter to smooth the signal, uses a peak detector to identify waveform extremes, and counts a step if the peak-to-peak value exceeds a threshold. The sensor features a dynamic internal threshold that adjusts based on step detection outcomes, improving accuracy during movement transitions and reducing false detections. A configurable debounce algorithm and a dedicated false-positive rejection block enhance step-count reliability, especially during walking or running. More information about it can be found on the IMU’s datasheet.
It’s not the first AI-enabled sensor around, and we had previously written about the Bosch Sensortec BHI260AP AI smart sensor with a 6-axis IMU and a built-in 32-bit customizable MCU, as well as the ST1VAFE3BX AI biosensor for biopotential signal monitoring and motion tracking also leveraging STMicro’s MLC and FSM.
The LSM6DSV320X 6-axis AI IMU can be purchased directly from STMicro’s website for $9.31, or $5.03 per unit for a 500-piece order.
Debashis Das is a technical content writer and embedded engineer with over five years of experience in the industry. With expertise in Embedded C, PCB Design, and SEO optimization, he effectively blends difficult technical topics with clear communication
Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress. We also use affiliate links in articles to earn commissions if you make a purchase after clicking on those links.
