Low Power Radar Sensors and TinyML for Embedded Gesture Recognition and Non-Contact Vital Sign Monitoring

Watch a technical conference talk from the tinyML Summit 2023 exploring innovative applications of low-power radar sensors combined with TinyML for gesture recognition and vital sign monitoring. Discover how ETH Zurich researchers developed an efficient embedded system using Acconeer's short-range Doppler-radar technology. Learn about their novel architecture combining 2D CNN and TCN neural networks that achieves 92% accuracy on 11 hand gestures while maintaining a compact 91kB memory footprint. Explore the implementation on RISC-V and ARM Cortex-M processors, making it suitable for battery-powered wearable devices. Gain insights into recent developments using Transformer neural networks and their application in an earbud form factor. Additionally, examine how the same radar sensor technology can be utilized for non-contact vital sign monitoring, including respiration rate and heart rate measurements, through signal processing and TinyML techniques that run on microcontrollers with minimal memory requirements.