Coordenação de robô autônomo por meio de visão computacional

Abstract

The demand for mobile robotics applications has grown considerably in recent years, especially due to the advent of industry 4.0, in which one of the pillars is autonomous robotics, the subject of this research. In this context, regardless of nature or purpose, autonomous mobile robots must interact with the world to achieve their goals. The problem of navigation in the context of mobile robots can be considered one of the biggest challenges related to the field of robotics, in particular, the location of mobile robots in closed environments encounters several problems ranging from the error accumulated by the sensors installed in the robot to the fact of the constant changes that can occur in such environments. The use of computer vision as a navigation tool in robotics represents an interesting alternative for controlling the movement of a mobile robot. Currently, the computational vision plays a central role in the robotics area for a range of different tasks, such as: self-location, navigation, recognition and objects manipulation, objects tracking, social interaction between human and robot, imitation, among others. Therefore, this master's research proposes the development of a control center to assist navigation and location of mobile robots in closed environments using the global view technique and a D-FCM to control the decisions. The experimental results present tests of the behavior of the navigation central in detecting obstacles and send bypass commands to the robot. Some difficulties inherent to the application of this technique, such as intensity and reflection of light, for example were solved applying an initial calibration of the system. In the experiments performed it was demonstrated that the system developed behaved properly.