This project' goal is to design and construct a device capable of detecting when an elderly person falls over in the SSH and is in need of assistance (cannot get up on their own). The system will use RADAR data sent thorugh a continually operating neural network which has been trained to detect fall patterns. The device will then be able to provide emergency services with a phone call should the patient either ask verbally for help, or be unresponsive after a fall.
The sponsor for this project is Dr. Bing Chen, and its duration was 2023 - 2024.
Capstone Team Members:
- Silas Perry
- Bekah Nelson
- Edoe Houndjoe
- Cole Long
Abstract:
Falling poses significant health risks, emphasizing the critical need for effective response
mechanisms. Traditional solutions, such as wearable devices or monitoring systems, suffer
from limitations including user discomfort, forgetfulness, or dependence on external
devices. Addressing these shortcomings, this report proposes an autonomous fall detection
system integrated with self-contained hardware and software. By embedding a neural
network model, the system processes radar data to automatically detect fall events. Upon
detection, an emergency response protocol is initiated to notify pre-selected contacts. The
radar neural network is trained on one subject and is put in place as a proof of concept for
such a system. Accuracy and robustness are critical for a device which will be in the
business of saving lives. Results show the feasibility of further implementation.
Recommendations include improving accuracy by training the network with broader data
sets and condensing the package size for consumer usage.
Read More
Contact Dr. Chen for full permissions to read more about the Radar Based Fall Detection project, including a full definition and proposal.