Robô explorador térmico de ambientes utilizando rede neural e algoritmo bug

Abstract

Autonomous navigation requires that artificial agents be able to understand and adapt to the environment, especially in the action of both static and dynamic obstacle diversions. The explorer robot senses the surrounding environment and learns from successful exploration experiences to plot the best routes and avoid obstacles. The present work describes an environment explorer robot that has a camera system and temperature sensors to identify thermal zones within the map explored. Because the robot aims to minimize manufacturing costs for possible destructive missions, the thermal image is comprised of a camera processing solution and a matrix thermal sensor. The robot is controlled by a Raspberry Pi model 3 b +, which has adequate processing power for the task, mainly due to the computational cost of camera image processing. Ultrasound sensors were responsible for "feeling" the environment, and autonomous locomotion is performed based on a combined solution of intelligent Bug algorithm and artificial neural networks. An interface accessed over the internet via IP was used to control the robot in both manual mode and automatic mode. Robot applications can be diverse, from fires to exploring inhospitable places.