Sistema Embarcado para Controle de Alimentação de Pets Baseado em Aprendizado de Máquina e IoT

Abstract

Brazil stands out globally in the pet industry, with significant revenue and a vast population of approximately 160 million animals, including 67.8 million dogs and 33.6 million cats. However, owners’ absence during travel or work periods can compromise proper pet care. In this context, technology, especially the Internet of Things (IoT), emerges as a solution to facilitate pet care. Many available solutions for pet monitoring and feeding have limitations, such as short identification range and high power consumption. Thus, there is a demand for more advanced and efficient devices capable of providing personalized care and handling power failures. This article presents the development of an intelligent system for pet care that combines IoT communication techniques and machine learning embedded on the ESP32 platform. The Smart Energy Supply System (SSEI) passed the 60-hour battery autonomy test. The pet identification system recognized the dog used as the model and served 5 meals during this testing period. In the neural network algorithm tests, two were examined: FOMO and YOLO, both correctly classifying the pet with accuracies of 75% and 83.3%, respectively.