ST integrates machine learning into automotive IMU
Based on ST’s proprietary MEMS technology, the ASM330LHHX combines a 3-axis accelerometer and a 3-axis rotary encoder (gyroscope) in a package measuring only 2.5 mm x 3 mm x 0.83 mm. The 6-axis module is used for motion and attitude sensing for functions such as vehicle location and digital stabilization solutions.
The ML core represents a hard-wired processing unit that handles AI algorithms directly in the sensor, ensuring extremely low latency between the detection of an event and the vehicle’s response. This, in turn, enables high-fidelity, real-time performance that requires significantly less system power and compute than solutions implemented with an application processor or cloud-based AI. To simplify application development, demo boards and libraries of free software samples are included. Available functions include vehicle standstill detection, attitude and heading reference, altitude estimation, towing and impact detection.
The ASM330LHHX offers two modes of operation, low-power mode for always-on applications such as telematics, anti-theft systems and motion-activated functions, and vibration monitoring and compensation. The current consumption in low-power mode with active accelerometer and rotation encoder is less than 800 µA. The other mode is the high-performance mode for application purposes that require maximum accuracy and minimum latency. These include precise location, V2X (Vehicle to Everything) communications, impact detection and impact reconstruction.
ST’s proven MEMS manufacturing processes ensure excellent sensor stability and low noise, as evidenced by the low measured Allan variance (AVAR) of the rotation rate encoder and accelerometer. The module also maintains its high accuracy level over the entire extended temperature range of -40 °C to +105 °C.
The ASM330LHHX is qualified to AEC-Q100 and is already in production. It features an LGA-14L land-grid array package with 14 pins and is available at prices starting at $10.44 (from 1,000 units).
Related articles:
Machine learning benchmarks for embedded systems
Machine learning predicts battery lifetimes
Machine learning improves ultrasound buttons
TI to add machine learning to codeless brushless motor drivers
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
