Due to the fact that elderly people have certain motor limitations, the chances of the occurrence of falls during the performance of daily activities are great and research shows that these falls can lead to some type of injury. Assistive technologies such as Wearable Computing allow independence, mobility and security for the elderly. Therefore, in this work, a wearable prototype was developed that made it possible to monitor the locomotion and fall detection of elderly people through a dashboard in Node-RED, where the prototype consists of an interface board between the devices and the wearable element chosen for the job was a fanny pack. The proposed solution corresponds to a system composed of accelerometer, gyroscope and camera, where the accelerometer and gyroscope identify the movements and the type of position of the user and together with the camera, a dual approach to fall detection is formed. In addition, the system also consists of an ESP32 and a Raspberry Pi model 3B, where the processing and analysis of sensor data were performed using a Threshold-based algorithm. The MQTT protocol was used to exchange data between system devices. The experimental evaluation was carried out with a user following the proposed methodology of performing a maximum of 3 repetitions per user for each possible position. The results obtained were satisfactory and the system obtained an accuracy of 60% considering all positions.
|
