Fall Detection Device Using Machine Learning on Embedded Systems
Siddarth Nandyala · 2024
This project introduces an Embedded Fall Detection System using machine learning algorithms for rapid response to protect individual safety.By employing advanced sensors like the gyroscope and accelerometer IMU with real-time data analysis capabilities, an alarm will sound immediately upon detection of falls by users as soon as a system alarm triggers instant alarms for them; notifications will then be sent out by both email and notification system to designated contacts.At its heart lies machine learning models trained on a fall dataset, which recognize patterns on efficient low-power embedded hardware to deliver discreet yet reliable daily life integration.When falls occur, immediate alarms provide user feedback, while email and notification alerts provide essential details that enable swift assistance from emergency services or caregivers.This project pioneers assistive technology by embedding intelligent fall detection using machine learning on low-power systems.This technology can increase accuracy and quickly adapt to evolving situations, ensuring timely interventions to mitigate fall injuries and enhance the safety of at-risk individuals. Background and Related Work ProblemFalling is the highest reason for hospitalization for older people in the United States.According to the National Institutes of Health, around 250,000 injuries and 11,000 deaths happen in the United States alone due to falling.During falls leading to severe injuries, receiving immediate help and assistance is vital and can be the difference between life and death.An estimated 9,000,000 citizens report falls annually, revealing a need for fall-detection devices (Verma et al., 2016). Current SolutionsCurrent solutions implemented include smart watches, necklaces, and more.These traditional methods of fall detection and alerts fall under a subset of a few disadvantages.