This project was submitted by R.M. Dreifuerst in Make With Mesh: The Cypress & Digi-Key Bluetooth 5 IoT Design Contest Sponsored by Embedded Computing Design and was one of the winners! We are now posting it in the community for all of our developers to see what great things can be developed with the CYW20819 - as well as leverage some of the material to get started on your own design. All project files submitted are attached to this community post. Check it out below!
Project Name: Elderly Fall Detection With Supervised Learning
This project leverages the CYW20819 Arduino Eval Kit to implement a machine learning powered automatic fall detection IoT device. Providing a fast and early response to medical emergencies is one of the most effective methods of ensuring recovery from hazardous situations - and so adding some automation to this process certainly brings to light some of the benefits in the medical world the IoT can bring. The project first created a data pipeline with the desired "fall" output using the on-board accelerometer on the CYW20819. The data was then fed into multiple machine learning models - eventually with a tested model being migrated to run on the CYW20819 processor to do automatic fall detection. In the end version of this project, if a fall is detected an emergency call on the user's smartphone would be initiated via Bluetooth. An image of the wearable device prototype, as well as some of the data collected is below.
All project files are attached to this community post.