Porting an AI-Powered Wearable Health Monitor to Zephyr on Open Hardware

Explore the process of porting an AI-powered wearable health monitor to Zephyr on open hardware in this 42-minute conference talk by Szymon Duchniewicz from Avanade and Jakub Duchniewicz from Tietoevry. Discover the challenges and decision-making involved in transitioning from a closed-source single-threaded system to an RTOS-based open-hardware solution with an AI model deployed on an FPGA. Learn valuable tips for getting started with Zephyr development, understand when and why to use an RTOS, and gain insights into best practices for deploying Machine Learning models on embedded FPGAs. Examine a project using QuickLogic's and Antmicro's QuickFeather board, powered by the Open Hardware EOS S3 System on Chip, which integrates TensorFlow Lite and an open-source FPGA toolchain. Discover how the device collects data from an SPO2 sensor, processes it with Zephyr, and utilizes an FPGA-deployed model for inference, with results displayed on an OLED screen. Gain knowledge about the open-source data used to train the model for blood pressure estimation based on SPO2 levels, specifically tailored for embedded scenarios.