Aplicação de internet das coisas na identificação e monitoramento de movimentos em exercícios físicos com pesos livres

Abstract

In a world increasingly integrated with digital media, a new technological communication resource has received more visibility: the Internet of Things (IoT). This feature consists of imbuing wireless sensors in devices used in the daily life of society to facilitate communication and explore the capacity of these tools in optimizing tasks. This paper integrates the IoT in the monitoring of physical activities with a focus on the use of free weights (weight plates), aiming to assist practitioners and coaches through an analysis of the performed movements and identifying exercises through the data read. This recognition was achieved through the use of machine learning algorithms such as decision trees, multi-layer perceptron, and kNN, trained to recognize patterns in the movement of weights. The plates were sensed with an ESP8266 microcontroller and an MPU-6050 inertial sensor calibrated with positional methods and digital filters. The microcontrollers send data to a server in the local network using socket communication. In the server, data are received, recorded, processed, and passed to the algorithms to accomplish the classification process which identified the exercises efficiently, achieving accuracies superior to 90% for the best cases. Therefore, this application brings great potential to make remote and asynchronous training efficiently possible, not invasive, and with a low cost.