TDK adds machine learning IMU to SmartBug
An all-in-one multi-sensor developed by TDK includes the ICM-456xy balanced gyro IMU (inertial measurement unit) and is machine learning enabled – to build, test, and deploy machine learning (ML) models on the IMU. A head tracking solution has been added for TWS (true wireless stereo) and augmented reality (AR) glasses.
The InvenSense SmartBug 2.0 builds on the original SmartBug, with added features for consumer and IoT applications. The smart, remote, data collection module for IoT will take digital transformation to the next level, said TDK, by not only being a trusted source of smart sensor data, but by providing the ability to create efficient machine learning models on a 2.5 x 3.0mm sensor, said Sahil Choudhary, product marketing director, at InvenSense, a TDK Group company.
The SmartBug 2.0 replaces the ICM-42688-P IMU with the latest BalancedGyro IMU, ICM-456xy. This TDK IMU is claimed to be the world’s lowest power six axis IMU (220 microA at 50Hz). The IMU allows developers to collect high quality IMU data for a much longer time, due to the extended battery life which is three times better than the original. The IMU also provides premium temperature stability and vibration rejection, making SmartBug 2.0 useful for products such as AR glasses, VR (virtual reality), optical image stabilisation (OIS), drones, mixed reality (MR), TWS, and robotics for prototyping, data collection, as well as machine learning-based algorithm development.
Developers, ODMs, OEMs, and product developers can now use SmartBug 2.0, along with the sensor inference framework (SIF) by TDK, to build machine learning solutions for the ICM-45686-S IMU. SIF software enables users to collect IMU sensor data, select custom features, build ML models, test performance, deploy, and run those models on the ICM-456xy IMU through the SmartBug 2.0. It allows users to develop custom motion classification algorithms and run them on a 2.5mm IMU at a current consumption as low as 30 microA. Examples include algorithms such as exercise classification (squats, jumping jacks, lateral raises, or push-ups) and wrist gesture classification (fight, clench, shake, or still mode). The small and wireless factor of the SmartBug 2.0 enables engineers to test these ML models in real use-cases and tune the ML parameters accordingly, to create robust ML models at the lowest power.
TDK’s VibeSense360 head tracking features allows TWS and audio companies to collect raw sensor and head orientation data at the lowest power (280 microA at 50Hz) and learn unique signatures about enabling spatial audio.
Another new feature is TDK’s Air Motion which adds swipe gestures, remote orientation gestures and accurate cursor tracking for smart TV remotes.
To improve data collection, the original SmartBug’s streaming and logging parameters have been updated to enable longer duration streaming and logging of multi-sensor and algorithm data.
The SmartBug 2.0 retains the user interface, Bluetooth Low Energy, Wi-Fi, USB and SD card logging together with asset monitoring, smart door lock and sensor fusion.
The SmartBug 2.0’s dimensions are 36.6 x 501 x 18mm.