Real-Time Fall Detection using ESP32 and AMG8833 Thermal Sensor: A Non-Wearable Approach for Enhanced Safety

Falls are a problem that primarily affects older people because they can cause permanent disabilities or even death. Efforts to prevent falls are highly desirable but challenging to achieve due to the inherent difficulty of eliminating all instances of falls. One approach is to use non-wearable sensors such as the AMG8833 infrared thermal sensor. Initially, set up the AMG8833 sensor in the appropriate position to capture the thermal data of the area where fall detection is desired. This can be done by mounting the sensor on a ceiling or wall, preferably at an elevated position with a wide field of view. Before using the sensor for fall detection, it's important to calibrate it to establish a baseline thermal signature of the environment. Continuously capture thermal data from the AMG8833 sensor in real-time. The sensor typically provides an array of temperature values corresponding to different pixels in its field of view. To develop a fall detection algorithm that analyses the acquired thermal data for fall events. The algorithm identifies sudden changes or anomalies in the thermal signature that may indicate a fall has occurred. To determine if a sudden, abrupt motion is detected by tracking the changes in the thermal data. Once a fall is detected, an appropriate alert or notification will be generated. This fall detection system has been built with an alarm system, mobile app, or any other form of communication to notify caregivers or emergency services.